A PB:rroSOCIOLOGICAL STWY OP' ml BOttœ:JtRif RUPURUII SAVADA1 BRITISH GUIANA.

A tbesis submitted to the Faculty of Graduate

Studies am Research in partial fUltil-

ment of tbe requirements for the degree of

Master of

Science.

by

R. J. A. Goodland

Botany Department, and 57 111gb Street, McGill University, Georgetown, Montreal, British Guiana. Canada • • August, 1964 1 TABLE OF CONTENTS

page

Title

Table of Cont,ents •••••••••••••••••••••••••••••••••••••••• o ••• o. iO 1 List of Figuree ••••••• 4 Acknowledgements •••••• ...... Introduction. • • • • • • • • • • • • • • • • • • • . • • • . • • • • . • . • • • • • • • • • . • • • • • • • • • • 8

Lite ra ture Review ••••••••••••••••••••••••• , ••••••••••••••• , ••••• 11 ' Description of the Region ••• ••••••••••••••••••••••••••••••• 16

Location •••••••••.••••••••• , •••••••••••••••••••••••••••••• lf

Geology ••••••••••••••••••••••••••••••••••••••••••••••••••• 16

Geomorphology ••••••••••••••••••••••••••••••••••••••••••••• l7

Climate ••••••••••••••••••••••••••••••••••••••••••••.•••••• lJ

Soils •••••••••••••••••••••••••••••••••••.••••••••••••••••• 20

Vegetation ...... 21

Ri parian Forest •••••••••••••••••••••••••••••••••••••• 22

Swamps ••••••••••••••••••••••, ••••••••••••••••••••••••• ?. 2

Bush Islands ...... 23

Marsl1es •••••• , ••••••••••••••••••••••••••••••••••••••• 23

Ac1ua tic Vegeta tion ...... 23

Mu.. ri Scrtltb ...... 24

Sa vanna •••••••••••••••••••••••••••••••••••••••••••••• 24

Ebini ••••••••••••••••••••••••••••••••••••••••••••••••••••• 30

Trinidad •••••• ·•••••••••••••••••••••••••••.•••••••••••••••• 30

Terminology ••••••••••••••••••••••••••••••••••••••••••••••••••••• 33

Fire •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• )6

Ani.rna.ls ••••••••••••••••••••••••••••••••••••••••••••••••••••••••• 41 2 TABLE QE CONTENTS (Continued)

Termites • •....•.••••••.••••••••• • • •... • • • • • · • • • • • • • • • • • • • • • • • • • • 43 Methods •••••••••••••••••••••••••••••••••••••••••.••••••••••••••• 46

Preparation•••••••••••••••••••••••••••••••••••••••••••••••46

Field Methods·-··•••••••••••••••••••••••••••••••••••••••••46 Selection of Stands•••••••••••••••••••••••••••••••••••••••47 Collection of Data••••••••••••••••••••••••••••••••••••••••48 Point- Quarter Method •••••••••••••••••••••••••••••••••••• 48

Soil••••••••••••••••••••••••••••••••••••••••••••••••••••••49

Treatment of Data•••••••••••••••••••••••••••••••••••••••••SO

Termit,es ••.•.•••.•••••••.••..•..•••.••...•••••••.••••• • ••• 51

f4a.trix • ••••.•.•••••••••••••...•..•..••••.••••••••••••••••• 51 Matrix: Summary of Procedure ••••••••••••••••••••••••••••• 56 Resulta ...... •...... •... ,...... ••.•...••.• 58 Introduction •••••••••••••••••••••••••••••••••••••••••••••• 58

Matrixs Nature of the Axes ••••••••••••••••••••••••••••••••67

Termites••••••••o•••••••••••••••••••••••••••••••••••••••••79 So1ls •••••••••••••••••••••••••••••••••••.•.••••••••••••••• 8) Floristics •••••••••••••••••••••••••••••••••••••••••••••••• 87

Floristic Analysis••••••••••••••••••••••••••••••••••••••••92 Physiognamic Analysis•••••••••••••••••••••••••••••••••••··93 Discussion •••••.•.•••••••••••.•••••••••••••••••••••••••••• 97

Clima.tic •.•.•••.•••••••••••••.•..•.....••....••.••••• 97

Biotic•••••••••••••••••••••••••••••••••••••••••••••••98

Edaphic ••• •••..•••••••••••...•••••.••.••••· ••••••••••• 98 3 ~ .9.!:: CON'i'ENTS ( Concluded) page

Appraisal of Methods ••••••••••••••••••••••••••••••••••••••••••• lOl

Further Research •••••••••.•••••••••••.••••••.•••.•.••.••••••••• lJ3

Post Script ...... •••••••.••.••••••.....•..••••••••••••.••••••• • 104

5 UJTli'na ry ...... • • • • • • ...... • . . • . • . . • . • . • • • • • • • • • • • . 1 C>f-J

Appendix•••••••••••••••••••••••••••••••••••••••••••••••••••••••l09

Stand Record Form •••••••••••••.•.••....•.•.•••••••••••••• l09

Presence and Frequency Form •••••••••••••••••••••••••••••• llO

Tree Data Form ••••••••.•..••..•••••...... •••••••••••••••• ll2

Tree Data Summary Fo~···•••••••••••••·••••••••••••••••••ll3 Matrix Form ••••••••••...••••••...•••••••••••••••••••••••• ll4

Plants of the Rupununi Savannas •••••• ..•...•••••••• • 116

References••••••••••••••••••••••••••••••••••••••~•·l30- 150 .4 FIGUru:5 page Tentative Sketch Map of sorne of the South American Savannas •••••lO

Climate at Lethem, Rupununi•••••••••••••••••••••••••••••••••••••l9

Illustrations: General View••••••••••••••••••••••••••••••·~···••28 Cerrado •••••••••••••••••••••••••••••••••••••••••• 28 Termitaria ••••••••••••••••••••••••••••••••••••••• 28 Muri Serub •••••••••••••••••.••••••.•••••••••••••• 28 Ité Swanq> .••••••• ci •••••••••• ••••••••••••••••••••• 29 Savanna Edge ••••••••••••••••••••••••••••••••••••• 29 Bush Island •••••••••••••••••••••••••••••••••••••• 29 Caeti •••••••••••••••••••••••••••••••••••••••••••• 29 After Fire••••••••••••••••••••••••••••••••••••••·29

Stand Location MaP••••••••••••••••••••••••••••••••••••••••••••••32 Distribution of Stands, 3 dimensions of Matrix Model •••••••••••• S5

Prevalent Species Table•••••••••••••••••••••••••••••••••••••••••6l

Main Tree Species Data Table••••••••••••••••••••••••••••••••••••~4 Graph of the Behaviour of 3 Tree Species •••••••••••••••••••••••• 6S Frequency- Moisture Graphs of Selected Species ••••••••••••••••• 66

Matrix Outline ••.•••••••••••••••••••••••.....•••.••••••••••••••• 73 Matrix Madel Graphs: 3 Minor Trees •••••••••••••••••••••••••••••• 74 Bulbostylis and Stenophyllus ••••••••••••••••••••• 75 Byrsonima crassifolia •••••••••••••••••••••••••••• ?S Curatella americana •••••••••••••••••••••••••••••• ?S Bowd1chia virgillo!des ••••••••••••••••••••••••••• ?6 Meaosetum loliiforrne ••••••••••••••••••••••••••••• 76 Stand Description •••••••..••.•••••••••••••••••••• 76 Trees per Acre (Density) ••••••••••••••••••••••••• 77 Termites and Sail Nutrients •••••.•••••••••••••••• ?8 Water Retaining Capacity of the Soil ••••••••••••• 78

Matrix Moàel Pictures ••••••••••••••••••••••••••••••••••••••••••• ~o Table of Environmental Characteristics •••••••••••••••••••••••••• 86

The Occurrence of the Main Rupununi Trees and Shrubs •••••••••••• 94 The Occurrence of Some Rupununi Plants •••••••••••••••••••••••••• 95

Key to the 2 Preceding TablesQ•q••••••••••••••••••••••••••••••••96 5 ACI:NOWLEG!:MEKTS

In spite of tlle usual xenophobies of snalœs 1 bugs, beat am vast dis tances, man is everywhere tbe gree. test danger to ld.mself.

But the exchanging or the amenities of the city for a simpler existence

ma.kes oœ especially susceptible to belp:t'Ul.ness am hOspi tali ty of which

tbere is no sbortage in the Rupunu:ni. I am glad to be able to aclmow•

led.ge these trieDde who bave made this ta.sk immeasurably easier:

Tbe entire am ubiquitous Melville tamily1 especially Edwina.

Mr. Brœst Bard.y 1 Mr. Geofi"rey Lomas am the ir familles, of tbe Rupununi Arma, Letbem.

Tbe Reve rem Canon Jack Bolden and Miss Elsie Schadde 1 St. Mary• a 1

Yupu.kari 1 and Annai.

Fatber Quigley S .J. , of St. Ignatius and Karasabai.

Mr. am Mra .Harry Bart of Pirara aJJd Meri tiziero. J ' Mr. Leonard D•Aguiar, Mr. L.B. McTurk, Mr. Caesar Gorinsk;y, Mr. Roy

Hewson and Dr. S.P. Legg.

The Police force aJJd the Pilots or the British Guiana Airways.

I was very tortunate in having the company of Nr. R. Persaud, Curator or the Je.nman Berbarium Georgetown, tor two weeks, whose lmow- 1 . ledge or the Guiana Flora is only su.rpassed by his indomitable spirit.

The Directors and Staff of the Govermental Departments of Agriculture, ll'orestry 1 am Geology am Mr. Ram Singb1 Curator of tbe Georgetown

Museum, were always ready wi th information and equipment.

In Trinidad1 Pro:tessor J .w. Purseglove of the Botany Department, University of the West lndies, President of the Association for Tropical 6 Biology, p.1t his excellent Berbarium at nay disposa!. Tbanlœ tor

exceptionally able help 1n tbe f'ield are due to the Curator, )te. M. , ' Bhorai. I should alao lilœ to thank Dr. Ahmad and Mr. Baynes •

Witbœ.t the belp of Dr. Thomas Soderstrom of t'be Smitbsœian

Institution, Wasbinston, tbis study vould have been useless. I should

lib to tbank most sincerely Dr. Soderstrom, Mr. c.v. Morton, ~Dr.

David LeUinger, au! Dr. Lyman Smith tor d.etel'llini.Dg rrry specimens of

Grass 1 Lover Planta am Pbanerogams respectively. I also tbanlt Mt-. W.A. Sands aD! Dr. w.v. Barris of tbe British Museum f'or identi-

tying the temites.

I should lilœ to tllank Dr. G.P. Patil &Di Jlr. Richard Pollard

of' McGiU UniYersity tor mathematical clsriticatiODSi Mr. GaVin Daly

and Mr. Jobn Lambert of' the Botany Depârtment f'or t'Deir interest,

CODBtructive sugsestions and proot readingj and Mr. S.G. Harrison,

Iteeper of the Carditt Museum, and Miss Violet Graham f'or taxcxlomic

balp...... This. study 1s largely a compilstive vork, but the taak of'

compilation bas been l18de easier am. more pleasant by the help of

Mrs. P. MacMorris of the lfev York Botanical Gardens Library 1 Monsieur

Richard Pu-é of the Université de Laval and Miss R. Maybev of the

Botany Library, McG1U University. A special mention must be made

to the staff' of tbe McGiU Inter-Library Loan Service to whom no

references are obscure, merel.y incorrect.

Profttssor Theo. Bills honored me by ~rship in his

McGill University Savanna Research Project ot 'Wbich this study is a smaU pert. I abou.ld l1lœ to thank him very mcb for a Geograpby

Departaient Research Fellovsbip vhich made this study possible, a1ld 7

tor bis farsigb.tëd and tl.exible attitude. I also tbank his students 1

my colleagues 1 M.J. Jaen anà. tf .K.P. Sinha • • I am deeply indebted to my Tutor, Proteaaor Paul Maycock, not OD.ly for his patieAce 'fini ready assistance both in tbe field aD1 ill the Labora tory but tor any la,low'ledge of lcology I may possess

8lld especial.ly for his exemple and vision. For this and for a UD.1vers1ty

Demonstratorsbip in Botany I o:tter my thanks.

Final.ly 1 . I talœ great pleasure in record1ng my sincere thanks: to my parents tor tbeir k.iDiœss, encouragement and interest. 8

IlfTBODUCTIOil

Savaœas are one of tbe largest aDl least lalcwn wgetatiœ ' types on the •vface of the earth. They are so poor~ 1rnown tbat

tbey bave œitber been adequately' mapped nor bas their area been

measured. BcMrver, the total area ot sava.unas in tbe world may

be almost 1 million square mUes • In South America 1 tbey caver

an area probab~ larger than an;y otber type of wgetatian on the

continent. SaV&DnBs occupy over 2fil, of the area of Brazil, as

mch as 100,000 square miles in Venezuela, ami not mucb lesa in

Bol11'1a &al Columbia (vide Sbantz, 1954) •

Althouf#l savannas are larger in extent than tbe Tropical Bain ' Fores ta 1 they have attracted lesa attention. Vith the accelerating popul.atiœ increase in the world, savaœas will become increasiDgly

important for tood production am settlement. It is therefo:re

œcessar.r tbat the savannas of tl:le vorld are caretu.l.ly stud1ed to

ascertain their ecolog1cal structure ani productive potential and to ensure tbe1r wise use.

This study concerns OlllJ a part of oœ small savarma aDl a

cursory aompart.son vith two others. It IIIUSt be empb.asised that any

resu1ts and statements bere reter onl.y to the area studied 8Dd are

oot uniwrsal.ly applicable. Apologies are ot:tered tor tvo orrmissions •

Pirst~, a sectiœ on the origin and causes of savamla bas been om­ mitted. as tbese aspects were not studied.. SecOD:i.ly, althcugh it is hoped that thiB stwly bfs been pu.t in cont.ext, the extrapolation of

1 ts resulte to other areas bas been scrupulous~ avoid.ed, consequeàtly leading to s.. incœclusiveœss. lo apologies are œeded for poeing 9 more questions than bave been solved.

Tbe aim ot this study is threef'ol4. Firatly1 as the title

1Dlicatea1 1t ia a pbytoaoc1ological atudy of' a savanna.' Although most attention bas been directed towarda tbe vegetation, animale &Di environment 1 as prime compoœnta of' the ecos ys tem are treated 1n some detail. Tbe Rupuœni savanœa are so intrinsically interesting, that digressions away trom the topic of' the vegetation are made wbere appropria te.

Secoml.y, this 1s one of the few instances where the quanti­ tative ecological methods of' the Wiscoosin achool have been applied to tbe tropics. It 1s thougbt tbat vith modification, they will p:rove to be mont uae:fUl than other methoda hi therto applied.

Thirdly 1 this study has been a part of a research training for tropical ecology 1n general, am pbytosociology 1n particùlar. ------,10 TENTATIVE SKETCH MAP OF SOME OF

THE SOUTH AMERICAN SAVANNAS

ClBA ~~ISPANIOLA HONDURAS ~

After& Dansereau 1948 Denevan 1964 Rizz.ini 1963 .. L.B.Smith 194.5 .· : Sca1e 1c4o,ooo,ooo. f (Gal1s Stereographie) 11

kiTBRATURE RBVIIW

It 1a dU'ficult to convey the state of botaey in the trop1cs

to anyooe unfam1liar vith tbe literature. Tbe ad.ded d1fficult1es

vh1ch atter.d vork in the trop1cs are to a certain extent outweighed

by i ta virgin nature and unl1m1 ted. scope • Corner ( 1946) vrites:

"To bring tropical bo~ to the present level of tempera te botany,

tbere 1s œeded yet 1n the tropics a hwlired times as ~ botanis ts

as tbe vorl4 has •ever la:lown." .Advice 1s rarely lacking but often

frustratingly atultif'ying. It ranges from the pragmatics 'Drav every­

thing, photograph everything, note everything" (A.H. Cburch in Corner,

1946) to tbe almost unobtainable 1 "knov your plants ••• at the outset".

(Beva 1n Tansley & Chipp, 1926). The former advice gives rise to the

tamiliar surprise at the lmowledge of tropical Foreaters etc. Uncom­ municated info1'1118tion, bowever profQUM., 1s usel.ess. Tbe latter ad vice may convert a fev dou~ty souls to taxonomy, but deters many. 'lbe abillty to distingud.sh taxa is more valuable than the ab1U ty to recall names • ICeying planta cu never be new to botany but plants can be studied vith a field name or number am. determined by experts later.

It taxanomy should precede ecology 1 tbere would be no trOpical ecology yet. Logic must occasionally be lett to logicians so that tbe simplest is studied betore tbe richest as Van Steenis advises. (19371n Richards 1952).

Tbe literature 1n general 1s extensive, although much is hard to obtain. The Interlibrary Loan Scheme bas be come an important re search as set, but lll&!lY magazines are non-circulating, as are older works, and 12 photocopying is expensive. Tropical South American botany ia still in the exploratory stage started by the great Naturalist-Travellers in the lBoo' s.

The ir vorlœ are often the beat or only etudies available. The progress of Tropical South American botany bas been recozded by

Gleason (1932a). Of the earzy Natural1sts 1 Hwnboldt am Bonpl.aDi,

(1819) 1 Schimper (1888} 1 Belt (1888), Sachs (1879) 1 Martius (18llo) 1

Spruce ( 1908) 1 Wal.l.ace ( 1878) am Batee ( 1.863) remain authorities.

Savannas have never been studied so ably or 1n so much detail since

Warming ( 1892) vrote on a Brazilian Campo Cerrad.o at Lagoa Santa.

Fanshawe and Swabey (1948) have wrttten about the Botaniste of British Guiana. The table swmnit of Mount Roraima (9094 ft.) attracted Botaniste from the time of Rmge (l8o5) and Meyer (1818) to the 1963-64 Bangor University expedition to Ron.imas sister peak

Kulœnaam. Tbe Schœburgk brothers ad vaneed tbe lmowledge of the Guiana flora ccmsiderably by the ir extensive travels and ass:lii I10U8 collectiDC,•

(18481 1876, 1922, and vith Bentham 1839 and Boolœr 181fo). They vrote viv1d and accurate descriptions of the Rupununi savannas • Other

Botaniste 1 auch as Appun ( 1871) and Iudhurn ( 1888), were less prol1f1c 1 tbough invaluable. .Jenman vrote tantalleingly briet descriptions of tbe savannas of Guiana in 1882 and 1888.

More recently, the botany of British Guiana bas been studied by tbe FOl'estry Department, (Am.erson, BQbanld.rk1 in Fanshawe and Swabey

1948), by the Geological Sune y, (Bartl.ett, Stockdal.e) am even the

Veter1Dary Service 1 (Praser 1946). William Beebe ( 1925) establisbed a researcb station 1n British Guiana a:f'ter the Great War 1 g1v1ng rise to Grabam's pioneer Flora (1933) am attracting the f'amous Botaniste 13 Gleason (1932) and Bitchceck (1936). b Oxford University Expedition

Society was the starting point of Richards Bcologi.ca1 etudies 1 {Davis

& Ricbards, 1933, 1934, ltichards, 1952) and Samvi ths ( 1929) taxonomie interests.

The I\lpunun1 aavanœs vere the subject of the excellent observa­

tions ot ~ra (1936). This work 1a compared witb otber sava.œa etudies

by Beaxd (1953). Tbese savannas are also being studied :trom different

aspects by other McGill University expeditions, sueb as Waddell (1963)

on the Anthropic factor; othe~ on se1l.s, Geomorpbology and Cl1ma.te.

There are tev acceunts of savanna ecology. The Suriname schoo1

umer Lanjouw (1936, l954) produ.oes tbe "Vegetation of Suriname", per1oi-

1cally. Beyligera (1963) &.Di Dœselaar (1964) bave presented phyto-

sociologica1 etudies af'ter the Braun-Blanquet system on tbe Intermediate

type of savannas in SuriDI!IBie. .Adjanoboun ( 1936) used tbese methails cm

savamas in the Lover Ivory Coast. Blydenstein (1962) applied Borth•

American phytosociological techniques to the L] anos. Theae tour papera

indicate the quiclœning pulse of savanna researcb.

Otber savama vork 1s being dœe in Braz11 and bas been excellen~ documented 1n "S1mpos1o s.~re o Cerrado", produced by the University of

Sao Paulo in 1963, am reviewed by Ferri, ( 1963) • The Brazillatt such studies savannaa are being studied frœ many aspects 1 auljare based on appreciable

:fieUI. vork ~i 1\b lesa untCJWiled speculation and extrapolation tban bas preViou.aly bedevUled the savanna controversy.

Veœzuela 1s the second center, atter Braz11, of savantl8 researcb in South America. The vork ot Lasser (1955), Ram:La (l959a),

Tamayo (1956), and Ar1stegu1eta (1959) is retemed to, c:oncem1Dg tbe - problems of the Llanos. 6ome of their vork 1s summariaed in "Saba:oas", 14 volume 3 of tbe IUnistry of Agricultures publication, "Reconocimtento

Agropecuano'', ( 1961) • Te.lœuchi, ( 1960 a • b) wrlœd in tbe Rio

Braœo sa'VUU:IA betveen the Ll&D08 Gd the Rupmuni.

Trirddad hu l.GDg been a center of tropical bo~, tormerly from

the Imperial College of Trepical Agriculture, nov tbe Univel"Bity of tbe ' West ID:liea. Tbe.. JI08t recent study is that ot Rlcba:Msœ ( 1963) on the Aripo aavanna. Be ccapans his observations wtth those of Marshall

( 1934) 1 )(vera ( 1933) am Beal'd. { 1946, 1953, 1.955) • Caywnœ 1 . (French

Ouiana) bas not produced ~ aaftDD.a atuilles. Lanjouw, (1936) reported

that the P'rencb Gu1aœ sa'Y1UlllU vere similar to thoee he studied 1n

Suriname. Bemoist ( 1925) brie~ described the French Guiana savanœs 1 au:l was tollowed much later by Rue (1958) ani Booek (l$60).

Taxonomie stu!ies haw been publ1Sbed by Maguire (19){8, 1953 tt.)

of the actbe lhw York Botanical Ga;rdens. Harrison, (1958) from trew,

pr

saT8DZ18, that of Ebini. Rizz1n1 ( 1963) and Boehne ( 1939, 19140, tt.)

are working on the Brazilian l'lora &Dl Pitt1er ( 1926, 1947) on the Venezuel.aD Flora.

Tbe 11terature on savanœs other than in South America bas

not been couulted in detaU. Apart from .AdJanoibœn ( J.962) ot' the

Ivory Cout, IDOSt Atrican stu41es are descriptive, bu.t hard to compare.

A tew Botaniste haw stu41ed both South Arœncan and Atrican or otber save.zmas. Cbevalier, (1925, 1928, 1929, 1931, 1.933), Schnell (1945, 1.952), Aubreville, (1947, 1949, 1953, 1961, 1962), Tt-oll (1936, 1950,

1956, 1959), Cola (1958, 1959 1 1960, 1963), aa:l Lauer (1952) bave worked on savarmas œ both aides of the Atlantic.

A sc:w::td aD! rapid entry into the 1iterat\U."e 1s attol'ded by 15 Verdoorn (1945) (vide Sllitb, A.C. alld S.itb L.B.) irrto Latin Allel"ican

Vegetation, by Roseveare (1948) into Latin laeriean Graael.aDls, 8Dd by tbe Silllpoeio sobre o Cerrado ( 1.963) into savannas • Tbese three vorks contain excellent bibliograpbiea which ramit.Y to most p.;Lblicationa •

'l'bis can be extemed to nrl4 saYUDBB by reference to Bills (1960) and Beard ( 1953) • Tbe lite rature of eaah topic is reviewed at the beginning ot the appropriate section.

' 16 DBSCRIPTIOI OF TBB REGION

:OOCATIOlf

Savannas were studie! in three areas of North Eastern South

America:

1. The l'ar-Interior sa vanna of the Rupununi 1 Brit.ish Guiana.

2 • Tbe In termediate savanna of Bbini, Bri tisb Guiana.

3. The Islam savannas of Trinidad. ( 0' Meara, Aripo am Piareo)

1. RUPUIIUJI

Tbe Rupununi savannas of South Western British Guiana are an extension of the vast Rio Branco savanna of North lastern Brazil.

The ltanuku mountains divide the Nortbern trom the Soutbern savanna.

Only the Nortbern Ru pmuni savanna 1 an area of about 2 1000 square miles, vas studied. It exten:ls from latitude 3° 30' to 4° North, and trom longitude 59° to 59° 4o• West (approximately). Tbe res­ pective geographieal bouniaries are the Kanuku mountains to the

South and the Palœraima mountains to the North, wi th the Rupununi river to the East ani the Ireng and Tekutu rivera of the Brazilian border to tbe West. Tbe administrative center, Lethem, is about

250 miles west of the Surinam border ani 110m11es s.s.E. ot the

Veœzuelan border.

Geolosy:

The Borthern RulUiluni savaona is part of tbe Precambrien shield exteniing :t'rom the Orinoco to the Palaeozoie Geosyneline of tbe Amazon basin. (McConnell, 1961). The Nortbern savanna may

' depth of 17 have been eroded to a,k,ooo :f't. by a vast east·west flow into ~e

Caribbean ··leaving the Pakaraimas am Kanukus elevated above the plateau. The present under1ytng rocks 1 designated tbe Talcutu formation, may be a series of red am grey shales am sandstones of the mesozoic (McConnell, 1961) wbich have occasionally been found as outcrops in the Takutu, RupunUD.i ~td Riva River banka.

(Bleakley 1 1957). This formation is presently overlain by deep quaternary sediments of sands am clays, vi th fluviatile and poeaibly lacustrine depoei ts. (Stark et al, 1959). · The tev

:t'lat topped bills am ridges may be former river terraces or lake beacbes. (Bleakley, 1957}.

Tbe Pakaraima mountains to the north are composed of pre- cambrian rhyolite• vith granitic intrusions and gabbroic and dolemitic dylœs • Tbey are partially coYe red v1 th quartzitic satds tones • The

Kanu.ku mO\mtains to the South are composed of a granitic gneiss ex­ teming into the Southern savennes. (Barran, 1956, Loxtan et al, 1958). Differences in geology have been sugsested as the causes of savannas by Waibel (1948).

Oeomorpholos.y:

The lorthern Rupununi savanna is on the north-south watersbed betveen the Amazon (Atlantic) and Essequibo (Caribbean) Rivers. Most of the drainage is into the Ireng-Takutu-Rio Branco system but the western side drains into the Rupununi-lssequibo system. Formerly, most of tbe drainage was to the west am there is evidence that the

Tali:Utu and lrezsg rivers joined the Rupununi river. Nov the Amazon system is actively capturing the drainage posaibl.y due to a slow ves tward tilting of this part of the shieJd. ( Loxton et al, 1958) • 18 The present divide 1s the hinge l1.œ.

To the north lies a large shallow depressiœ from Gocd Hope to

Armai at the base of the Pakaraimas which rlse s tee ply upwards • This

depression, the lforthern Flats 1 1s flooded in the wet season joining

the Ireng am Rupu.mmi· Rivera. The Kanuku mountains are skirted by ;• a sllghtly elevated erosion sur:f'ace (Stark, 1959) exteming to tbe

Borthern Flats. This raised surface is composed of Wldulating la teri tic

grave1 ridges on s8lld am permits tree drainae,e. The shallow depres­

sions between ridges however are less well drained. In addition the

lforthern Flats are I)oorly drained. Defective drainage bas been naœd

as a cause of savannas by Bennett and Allison (1928), · Michelmore (1939) 1 Beard (1914). Cole (1963) suggests that differences in Geomorphology

lead to the varying distribution of savannas.

The Horthern Rupununi savanna is in a state of "hydrological

1mbalancen (Loxton et al, 1958). Many of the rivera are dry in the dry season but may rise 4o feet to flood vast areas every wet season.

The savanna 1s only about 350 feet above sea level but most of the

· va ter travels over 2 1000 miles to the sea.

Clime.te:

The weather in' the Rupununi savannas is pleasant in the dry

season am tolerable 1n the wet season. As 1s usual for the tropics

the di umal temperature variation ( 10° F.) is greater than the annual 0 variation ( .,0 F.). The mean value in the Rupununi area is 83 F.

The Relative Hu.m.idity remains fairly constant nocturnally at 8()1,,

falling to 101> in the wet season day am 55~ in the dry season day.

During the short but marlœd. wet seasœ from May to August

about 13 in. of ra in falls eaeh mon th am this total is ~ o:r the 19 CL 1MATE A T LET HEM , RU PU NU N 1 .

'

15 11 15k.

10"

TEMP. 80'f.

J F M J J A s 0 N 10 6 6 16 23 22 28 23 7 5 4 55 51 53 58 66 69 70 73 52 52 52 55c a= month b= no. rainy do ys c= rel. humidity% at 1400 hrs. meon roin =70·06". mean temp. = 82:Tf. (7 8·4 am -87·0pm ). mean wmd 7·8 knot s. 20 amual precipitation. In tbe long dry seasœ :f'rom September to April

œ].y 1 in. or 2 in. of rain falls each mœ th.

'l'be vi.Dl speed increaaes gradual].y in the lfev Year from 2 mots

to 6 mots in the vet season (May-August), to the atrong vi.Dls of 15

knots af'ter the rains.

This cllœte is not ideal for plant growth. 'l'be rain is

unevenly dis tributed 1 am tloois altema te vith drought. The

effective raintall is much lesa that the actual rainfall of 70 in.

Tbe saaly nature of the soils, the ir poor vater retention and the

high viais aftord very dry ecological co:adi tions. The •lternating

of tbe wet am dry seasons has been postulated as a cause of savannas 1 vide Troll ( 1956) •

Soils:

1he soils of the Rupununi savannas have been studied many

times (P'ollett Smith and Prampton, 1935, Duthie, 1939). The most

recent vorlœrs using samples 1 surveys am air pbotographs produced large scale, tentative soil maps of most of the area. (Loxton, et

al, 1958, Stark et al, 1959). It bas been suggested that the RuPJ.nuni soils were weatbered umer bumid conditions, vere covered deep].y by wh1 te saDi wbicb vas • subsequentl;y lost, am are now re-exposed. (Loxtan et al, 1958).

Tbe Rupununi soils have been classitied on the basie ot sample profiles am areas into many different types.(Loxtœ et al, 1958, Stark et al, 1959). However, the soils may be classified into three main intergrading groupa:

1. Gravell;y soils: tbese occur œ upl.and sites and are compoaed

of secondary laterite or perdigon (Bemett and Allisœ, 1~) 21 ranging from 3 in. dieme ter pebbles to pisoll ths w1 th varying

amowts of red. eartb becoming paler below.

2. Grey soils: oecur mainly œ uplatd sites atd are composed

of grey sand vith silt of silty clays. Massive or primary

laterite occasionally outcrops in both these soils.

3. Soils of the f'lats: these are pale colored sandy ailts and

clays occasionally wtth a thin organic cruet. They may

become yellowish below and may contain red iron or black

mangenese ncdules • Tbese soils are otten flooded. The

fluctua ting water table promo tes lateri tization.

1 In general the soils are bigbly acidic and very law' 1n all nutrients. Tbe cbemical weatbering of tropical soils 1s tbrice as fast as temperate weatbering and leads to very deep senile soils •

I:f the sur:f'ace o:f the soil is penetrable, as on lateri tic ridges wbere the drainage is rapid, tben tbe bigb rain:fall and acidity lead to deep chemical weathering. The excessive elluviatiœ of organic matter and clay reduces the water retaining capacity of the surface soil. The resulting deep desiccatian turtber increases loss ot soil cons tituents and aggravates the cycle. If the sur:face becomes compacted by the inte:dsity o:f the rain, bigh insolation and the re moval of the vegetation by tire, sheet wash, gullying and serioo.a erosion usual.ly tollowa. Bulrl18 is rapidly decompoaed in tbe tropics due to the activity o:f the soil microflora.

Vegetation:

b Jforthern Rupunun1 savanna is bounded on tbree aides by 22 forest am to the west by tbe Rio Branco savanna. T.be Pe.k:araimas 1952 to the north support Dry Evergreen forest (Fansbaw) and Deciduous e. 1936 Monsoon forest (Jfyers). The Kanuku.s to the scuth support soœ

distuzlbed Selli-Deciduous ()(yers, 1936), Montane (Fanshawe, 1952)

torest. Tbere is more interdig1te. ting savanna in the Palœraimas

than in the Eanukus •

Tbe vegetation of the area 1s not uniform. The f'ollowing

vere distinguished:

1. Riparian Forest

2. Ité swamps 3· Bush Isla.Dls 4. Mars .b. es 5· Aquatic vegetation ,.. o. Muri scrub 1· Sa vanna

0~ tbe saV&Dna was studied in detail.

1. Riparian Forest:

Ripar1an Forest e.xteois from the mountains along creeks.

It is caaposed ot tall forest trees resembling monsoœ forest (Myers,

1936) vith Palma aDi Bamboo wllile the forest floor bears a dry 11 tter.

2. Swamps:

'lbe Ité palm (Mauritia m1nor) borders seme smaller creeks

and my ind1cate areas of' continuous groum:water mowment or clay

substrate (Jenma.n, 1888). Rank grasses aDl sbrubs especially

melos toues occur between the :çalms • Tbese palms are much used tor 23 tbatch 1 ropes 1 and f'cxlder aD1 are becoming lees :f'requent. Mauritia

setigera occurs in• the Aripo sa'VaDD.a and M. vinitera in the Amazœian savaimas.

3. Bush Isltmda: .

A tew isolated patches of dry forest c.alled Bush Islands occur

in the area. 'l'bey ra.np from 50 yards to 3 miles in diameter am

are eoœpœed ot tall forest trees particularq o'f tbe M;yrtaeeae and

1 1 Leguminosae (~ers 1 1936) • '!be ir oceurence bas been !8l'1ous]3 attributed to:

(a) Bain path f'rom ltanukus (~therford persona! coamunication}

(b) :rossil laterite as an aquiter (Sinba persona! communication)'

(e) unimpeded drainage (Stark, 1959)

(d) AbaDioœd dwelliDg leaviDg tru.it trees in a dung rich plot

.(Weddell, 1963)

(e) Birds that drop :truit seeds wbile perched in a tree (~rs 1 1936)

4. Marshes:

'l'beee are lov areas that remain fiooded for a lœger seas01138l

period. tban sedge meadows ani support a mixed t'lora of aquatics 1 some

su:f'f'nlticose herbe am shrubs. ~ marshes become des1eeated in the

dry season.

5. Aquatic Vegetationl

Deep pools ha~ woody species only around the ir edges.

Booted or t'loating ~othYtes am aquatics oeeur alt.hou.gb most

pools dry up seasœally. 24 6. lfur1. Scrub: e A peouliar am intereeting type of vegetation called Muri sorub troœ the verœcular name tor tbe dœinant shNb Jlumiria spp.,

vas obserwd in tvo small separate areas (Tivivid aDi lbini) whioh

beoomes extensive in Suriname aJil other parts of British Guiana.

:lt ocours exolusively on white •arrl and consistls of circular clumpJ

ot bushes aD1 small trees separated by bare "paths" ot wh1 te sand

about 8 tt. Yide. In the clumps there are a tew forbs, grasses and

sedges. The :nora is extremely oharacteristic and apparently very

similar in Suriname. fibri Scrub vas not quantitative ].y sampled but

a presence list vas compiled:

Woodz Plfgts

Byrsœ.ima spp. Axonopus lœitulœns is Clusia œmorosa Cassiatetraphylla Bum1r1a balsamitera Catasetum Bumiria floribunda Comolia villosa Licania incana Epidendrum Myrsinaceae Legenocarpus weigelti1 Ochnaceae spp. Schizaea spp. Ormosia costulata Syngonanthus umbe llatus Pagamea capitata TUlaDdsia f'lexuœa Retin1phyllum sohomburgk11 Xyris paraensis

7. Savanna:

Tbese savamas are dominated by grasses and. bave open grown

trees in them. Savanœs vary t'rom sites where sedges are co-dominant

vith grasses vith extreœly scattered trees, to those wbere tbe trees

to:rmed a canopy of almost r::t:fl,. Sbrubs and forbs are usually present.

The dominant grasses are Trachypogœ 21umosus api .Andro~

apgustatus occuring vith subsidiary grasses suoh as Alonop!.lE ghns1tes 25 Aristide ~ti tolla, am Mespeet.\lm Wiitpxp • More than a dozen species of PaspalWD have been reported tran tb.e savannas. The

grasses may grov over l meter tall am 1n bunches 15 - 30 ems. epart.

Otten almost be.l:t' hidden by the tutts of grasses are small but brightl;y colo:œd fotbs auch as Po:tvpla spp., Borreria S:Ql?• 1 Cassis spp., t»rrem~a spp., Ey,plyulus epp., St:ylosanthes spp. and Ruella sp. etc.

Sedges are always present and may becoœ co-dominant. They are Bulhœtyl1o spp. aDi FiQ!gristylis spp. on the dryer sites w1 th

BbyncbospOra spp., Stenophyllus sp. 1 Dichromeœ and Scleria spp. on ... the .mois ter sites • Depressions aDd :t'lats may support a sedge meadow w1 th some grasses and berbs, ani an occasional suftruticoae herb.

'1be :t'lats are inu.ndated for a shorter time than the marshes and also become desiccated in the dry season.

The most frequent tree is Curgtel,la americana, Yi tb BYrsonima crassif'olia. Bowdichia virgilioides and ttrt;e.ceae SR· While the tallest tree vas P1umer1a ingdqm, up to 10 mt., most trees were less t.han 4 mt • higb •

The shrubs and woody berbs are small but neither b~hy nor dense. Tbey include Pstdium SiR., Palicourea, Casearea, Randia,

8Dd Bl'Y):l!r9!Ylum spp.

The trees never form a complete canopy. Bowever seme plant species were found restricted. to the shad.e of an individual tree or • two or three trees together, often associated with a te:rmitartum, tb.e whole otten comprising a small clump. Such species are Amasonia,

~~ Dichrotpena, and Beliconia sp.

Most trees bad certain special features wbich are designated as characteristiet of savanna trees. Many plants had very large leaves. Sucb macropbyllous species include Curatella vith leaves up to 8 inches long, ani Pallcou.re§ vith rather longer leaves. Furtber, the leaves were otten sclerophyllous • Where the law nu trient level limita grovth, the accuw.lated. carbohydrate is converted. to cellulose as sclerenchyma. ibis has been called Oligotrophic scleromorphism (Arens, 1958) •

Saae saV'Billl8 species exbibit xeromorphic features such as a vaxy cuticle (Pluœeria) 1 hairy leaves (Byrs9J11œ verbascifolia), t.hick corlcy bark (Bol«lichia sp.) 1 low stature and x;ylopoiy (Rawitscher, 1948} ( Camarus incoeptus) • Tbe tree species are precocious in that they flower wb1le less than a meter tall. They are aèle to aprout :t'rom grou.nd level a:f"ter being eut or burned. Most savanna species are heliophytes ani some produce serotinous seeds.

The nature of the forest-savanna boun:lary has been the subject of a recent UNESOO colloqu1um, but vas not included in the present s tudy. In the Rupununi savanna the bow:dary vas often on a topo~ graphical feature such as a river. ihe abrupt nature of the boUIXiary may be due to t'ire • 'n'le savanna near the boundary is usually of the

orchard type. Altbougb gradations t'rom savanna to forest may exis t 1 they were not observed in the Rupmuni. Flor1stically, the savanna bears almost no relation to the forest.

As vould be expected., the aspects of the vegetatiœ. are changed by seasonali ty. In the wet seasan there are ~ ephemerals especiall:y hydrophytes • Perhaps most of tbe savanna species come into .f'lower soon af'ter the wet season especially the torbs. Some ot tbe trees may blossom 1n any Christmas ra in. Fev if' any of the

trees are deciduous 1 but many shed their l.eaves a:f'ter the passage of' 27 fire. Fire causes the grasses to produce succulent green shoots among the charred stubble 1 am sorne plants f'lower only a :t'ter fire (S;teaophyllus (= Bulbostxlls) paradaxa). 28 ILLUSTRATIONS

TOP: Foreground is a gentle slope of orchard savanna with part of a "Bush Island lt to the rigbt. Middle grouni shows a sedge meadow vegetation with seattered shrubs (Camp sujo) merging into an Ité swamp along a water eoune • BeyODl the swamp lie the Northern Savanna J.f'le.ts running to the Pakaraima Mountains.

SECO!I): Cerrado vegetation on pisolithie uplar.d • Antonia ovata

to the left1 Curatelle americana bush to the right.

T:a:IRD: An area of dense but rather sœll termitaria, with seattered shrubs and trees • ..

BOTI'OM: Muri acrub on white sam • The main plant in the sand 11 path" is L@s!nocarpus ve18!lti1. The sbrubs are Humiria s pp. etc. . 29 ILLUSTRATIOE

TOP: Campo Liaapo in trœt of a mature Ité swamp (Mau:ritia minor) vith ab.rubs • Ligb.t can be seen penetrating right through the swamp from the other side •

~OID; Sav&DD& edge 1 Campo Sujo on the flat f'oregrou.nd running up the slope to the lett. Other slopes bear forest.

miRO: Oœ of' the uncomman "Busb Islands". Note tbe abrupt ecotone betveen Campo Limpo aDd forest. Ifolit tbe Palms in the Bush Island.

rotJRTH: An area ot dense Cacti (Cenus ap.t) vith RaDUa (white f'lc:Ners) md otber ahrubs • The Bortbern Sa vanna Flata lie in the distance •.

BOT"roM: Orchard savama just atter tire. 'l'he bare grou.nd 1a showing although tbe ash bas not yet been waahed or blown avay. The tree leaves are dea4 and will taU but the trees are probably still allve. 30 2. EBDI

lbini 1s about 6o miles trom the coast up tbe Berbice River

and 70 reet above sea level. The soils are derived from tlle deep

coastal sediments and are brown and white, acidic sands. Tbe cllmate

is similar to the Rupwumi but vith a less severe dry season and only

parts are fiooded. llbini 1s surrounded by seaaona1 forest (hnshave 1

1952) 1 and supports a similar vegetation to the lllpununi. The soils

have been studied by Follett-Smitb (1930) 1 and Stark (1958). The

flora bas been described by Martyn (1931), and Harrison (1958).

3 • 'l'RDIDAD

1bree small savannas were studied 1n Trinidad. Aripo,

(locally designated nuoi>er tvo) on the Cumuto Road at Waller

:f'ield is a tl.at lOO acres or tine grey sand enclœed by seasonal

svamp forest. Apart from a small bush island, trees occur only

on the edges • It is fl.oadeà. 1n the wet season and 4esiccated in

the dry season. (Richardson, 1963).

Piarco savanna lies on the edge of the Intemational

Aerolraue 8.tld is much disturbed. 'lbe part studied vas well

treed ani bad deep hogwallowing. The soil ws an ac1d1c brown sand.

O'Meara savanna 1s on the Churchill-Roosevelt Bighwy v1tb1n

an industrial devel.opment scheme. This savaœa is siœilar to Piarco but not as wet nor as hogwallowed. The so11 1s a t1De brown sand.

The climate of tbese savannas is sii:Dilar to the Jupununi 31 although tbe rain:f'all ia sllghtly less at 10 in. per month in the wet season, sl.ishtly' more in the d:cy season (3-4 in.) with a higber aœual mean (lOO in.). The temperature is sliglltly higher, while the humidity 1s similar. Tbese savarmas are about 100 feet ab ove sea level. 32

STAND LOCATION MAP

"·.. MKARAIMA MOUNTAINS :·:llaraaaboi

BRAZIL .pi reno RIO • oOo 0 : 0 0 0 : 00 0 .• 0 0 ••• lethem 0 0 • BRITISH 0 0 o: 0 GU ANA o o .. . ·· oo f .·. · KANUKU .. 0 ••• 3'N.+ • m ocÙ't/'i. 60'W. TAKUTU RIVER BR AZI L

.village .... mountain edge o 1t a nd location river 33

There is contusiœ am. misUilderstand1ng eonœrning the term:iDOlogy of tbe savannas • This 1s not surprising for very little vork. bas been doœ on them, am althougb. savaœas may vary greatly in all aspects :from place to place, the conclusions dravn from oœ e.rea may be glibly extrapolated to account tor otber savannas.

This confUsion is mainly due to the use of dift'erent criteria, tbeir inconsistancy or lack of criteria. 'Ibis situation is aggravated by the use of local terminology am the dif'f:l.culties of la.nguage. It is not the aim of this paper to present a classification or terminology for savannas, but merely to clar11'y the terme used. ff01œnclatural hierarchies have been presented by Burtt-De.vy {1938} and Cole {1963). llowever, be fore any system is accepted, more detailed work is œeded to f1rJi the renge of variation of savannas •

The term se.vama is used in the sense of Sie:f'r1tz {1943) 1

Keay {1947), Curtis (1959}, and Bray {196<>). Savamas are plant camuunit1es dominated by grasses 1n vhich there are open grown, seattered trees. The grasses are termed dominant because they are the higbest, more or lese closed stratum. Sed.ges am torbs are than al:most e.lvays present, but usue.lly lese important/grasses. Curtis

(1959) proposes the arbitrary limita of more tban one tree per acre, but less than m ee.nopy. Beaxù (1953) :follows :Bew (1929) in allowing that savannas may be treeless • ~is author agrees vith

Eiten (1963) in that: "If its (savaxmas) meaning is so extetlded. that it includes everything from pure grasslani ••• to taU closed. 34

woodlams 1 then i t los es all meaning." Eiten ( 1963) wou.l.d use

the term savama woodlaDl f'or tbe 2 5 to 5r:fl, canopy class • This seems more realletic as even a 25i canopy was rarel,y encountered 1n the Rupununi savannas •

Cerrado is a Portuguese word mean1ng dense or cloSed, and includes everyt.hing f'rom woodlaais (cerradao) to gruslands (campo

llmpo). It was used Vith the word Campo (Portuguese f'or field) as

Campo cerrado, to distinguish areae vith scattered. woody vesetation

from areas without any woody vegetation (Campo limpo or clean field).

Tbe term Campo aujo {literall,y: dirty :field) ia used f'or an inter.. mediate type vith a very f'ev scattered, law shl"Ubs. Cerrado alone is used lesa in a pbysiognomic sense and more to ref'er to the distinc­ tive flora in general, wbile the term Cerradao ref'ers to a rare type of' tall closed woodlam, ·possibly vith some cerrado trees but vith dif'terent shrub and herb layers (Rizzini & Ber1nger, 1962). Occasion­ ally tbe vord savanna may be quali:f'ied by orcbard to dist1nguisb between areas w1 th more trees than others . Also s tudied in tbe Rupunu.ni we:re areas wi tb no trees • The se may be called sed.ge meadow, the betbaceous swamp of' Beard (1953), Patano herbacee in Spanish, Campo de Varzea in

Portuguese, or grassla.nd (Campo limpo, or campo sujo i:f' shrubs were present).

'l'.bere are savaxmas in South America where the trees present may be Palms (CoJ!micia, Sabal, Mauri:t.ia) auch areas being called

Palm &vannas {Beœett & Alllson:, ].928), otber in whicb pine trees

(l!mt! cf.r1baea .~m P. oocaœa) occur are called Piœ savannas (Parsons, lÎ55, Denevan, 1961, ani Jobannessen, 1963).

Most ot the ltupunun1 savanna vould be classit:l.ed by Trochain, 35 ( 1954), as savaœ arbustive, garrigue, and by Cole ( 1963) as savanna

parklaD1 but the grasse a·~ not be over 8o ems • nor Mesophytic.

!l'bey are classi:f1ed by Myers {1933) as 1ow bunch grass savanna, tbougb

the grasses may be t.all, ( over ib ems • ) and not particularly bunched.

Parts ot the Rupununi are 11ke the "Oak·:ridge" vegetation in Homuras

(Charter, 1941) am the Cbappara1 of the Llanos. (Ramia, 1959a, Bl.ydenstein, 1962). Chapparal is a usef\11 Spanish term meaning

savanna where the most trequent tree is the Chapparro. (Curatell!

amencana}. savanna At Ebini, British Guiana, there vere tew trees, and mostAms

ot the grasslarxl am sedge meadow type. Tbese are like the savannas described f'rom Suriname by Heyllgers (1963) ani Donselaar (1964),

from Cayenne;fianjouw, 1936) 8lld from Puerto Rico by Gleason ani Cook (1926). Aripo, Trinidad, is classi:f'ied by Richardson (1963) as having a grass sedge vegetation, by Marshall (1934) as a low grass herbaceous swamp, ani by Beard (1946) as JQirsh savanna subsequently renamed seasona1 swamp (1955). O'Meara am Piarco, Trinidad, savannas are Orchard sava.nnas (Marshall, 1934). 36

From observations, interviews am the literature 1t seems that most of tbe savanna stud1ed. is burned nearly every year.

Cœtrolled use of tire by earl.y man may bave started one million years ago (Stewart, 1956) and the" is eVidence tbat he used tire drives

10,000 years ago (Sauer, 1944). Hanno reports in 11 Periplus" the firing of African bush in 500 B.C. (Stewart, 1956). Se.uer (1944) thinks that man may bave fouDi savannas when be arrived in South America. The low incidence of animal endemism may ioUcate that tbe savannas are most recent (Batee 1 1948) •

Kubnholtz-Lordat (1939) reports on the world vide occurrence of

~rposet\ü burning and Stewart (1956) documenta incemiarism from Alaska to Tierra del J\lego. &\vers (1936) eupported by Ravi tacher (1948) tound evidence of more or less 11-equent burning on all the savannas he Visi ted in South America. Although Nyers (1933) bad seen the Aripo savannas,

Trinidad 1 Beard (1953) tou:œ no evidence of burniD.g. Tbese savannas have subsequent1y been burned (Ricbardson, 1963), in coamœ vi th other

Trinidad savannas ( Marsball1 1934) • There are many non-human agencies blamed tor causing t'ire, volcanoes for example (Bougb1 1926}, but vell documented cases of spontaneous tires are rare (Viosca, 1931, Lindemann, 1953).

Ligbtning as a cause bas been blarued by Weaver ( 1954) and Budowslti (1956, 1958), a:œ exonerated. by Plummer (1912), Braun-Blanquet (1932),

Glse.son ( 1913) and Stewart { 1956) •

Tbere is no doubt tbat nowadays man is the chief cause of tire.

The AmeriD:iians vere starting grass fires wbeD the Spaniards arrived 37 (Schmeid.er 1927, Batee 1948). This may have been to chase game {Sauer, 1944) includingturtles (Budowski, 1958) {lleyllgers, 1963) or the enemy (Berod.otus 447 B.C. in Stewart, 1956) or even to clear telled trees during shitting agriculture (Bartlett, 1956). Most of the tires Ut by the .Amerindians are for fun (M;yers, 1936) 1 some are accidenta! (Stewart, 1956) wbile some are for comunmicaticn

(Bartlett 1 1956).

Today tires are used to berd cattle (1Culmholtz•Lordat 1 1939) am even bees (Budowski, 1958) 1 8Di to provide the 1'1-esh grazing which sprouta after tire. (Myers, 1933, 1936). Fires Ut betore the grasses become tall, tibroua am mat~ are Ulœly to be lees dangerous to property thau accidental, intrequent tires (M;yers, 1936, McCorkle,

1952). Pires may reduce ticlœ (Boophil.ut 'llicroplua~ and other pesta.

Grass around dwellings is burned to reduce the snake hazard.

T'be incidence of tires has increaeed in the Rupununi due to the increase in population, am availabill ty of matches, am the change from hunting to pastoraUsm ani agriculture • That tire has a profound effect on the vegetation bas long been appreciated by pastora11sts (Ku.hnho1tz-Lordat, 1939, Beard, 1949).

The behavior of !ire has not been studied in detail, although the gross etf'ects are well known. J'ires start in the quickly drying grasslam am may pe:aetrate the forest, but rarely the converse.

Bowever though tires are rare in tropical foreste they are not

Ulllmawn. Tbe fores ta north o:r the Rupununi savanna have bumed tvice within living memory (McTurk, 1963, personal cOIIIIIUnicatiœ).

The high water table in Gsllery am Bush Islam forest is supposed to pntvent f1re (Bmowski, 1958) but both bum and disappear. The 38 Kanuku mOlmtains may burn to the sUIIID1 t every 10 years ( Gilliard, 1962) •

Grasslani burns 1n an irregular strip pattern. Usually only

tbe aeria1 parts burn, leaving undergroum parts a1ive. Such surface

tires are not tieree altbou.gh t.llere are very :tev etudies ot tire

temperatures (U.yvard, 1938, Beadle, 194o). Tbe uninterrupted

topograpby of tbe sava:ona am the long dry season tac111 tate the

spreed of such tires. 'Ibe eff'ect ot tire on savazma biomass bas been sUldied by Roland am Beydacker (1.963).

Geœrally tire changes forest to grassla.Dd or savama, as may

be seen in shi t'ting agriculture, but the grasslam may require buming

· tor 1ts perpetuatiœ. T.b.e reversion. ot grassl.aDd to forest when buming and sometimes grazing, is stopped, bas been well documented. (Gleason, 1932, .Aubreville, 1947, 1953, Scbnell, 1945, Richards, 1952).

'lbe increase of forest into annually burned grassland is rare (Eggellng, 1947, Jackson, 1956). Pire increases grass ani berbs at the expense

ot woody species (Weaver, 1954), and some plants have been uaed as ind1cators of burning (.Aristeguieta, 1959). Frequent buming increases

tbe bare grou:od (Curtis and Partch, 1948) unti1 tbe vagetatiœ is so

poor tbat tire carmot travel. Some savanna trees may ove tbeir sur-

rtval to the absence or combustible material in their Vicinity

(Budowsk.i 1 1958). PiNs early in the dry season are less serious

tban late tires (Bense1, 1923 aD1 Jackson, 1956).

J'ire may not al.ter tbe phyeica1 properties ot soU (Beadle, 194o) but erosion am losa of productivity occurs (Sbantz, 1947,

Chevalier, 1929, Budawski~ 1958, Sclme11, 1945) aDd surface run .. off increases (Anderson, 1949). Fire favors some graz:tng animale in Atr1ca (Budowski, 1958) but harms o:hers. Nearly every tree near 39

the line of fire is occupied by a bawk wai tiDg for a Uzard 1 snake or rodent to break cover. Al though fi re may not kill a tree 1 any wounds attord an entrance for insects aDi disease • Most of the ash is washed too deep to be recycled.

Sa vanna vegetation is well adapted to fire. Kuhnnol tz-Lordat

( 193~) coined tbe term P.frophyte to mean a plant that is :f'requently bu:rned away and may have more wood be law grouDi than above. Pyrophytes have other charaoteristics. Many pyrophytes are insulated by a thick oorky bark (Curatella, Bawdichia) am may be impregnated by a fire re• sistant substance (:Byrsonima) Ulœ the silica 1n the leaves of Curatella.

(Racbid-lldwazods, 1956). Woody monocotyledons bave no vascular cambium and are not killed to such a great ex.tent by fire. Many savanna plants

have seeds that are resistant to fire 1 and are often serotinou.s. Many are precocious aDi can flower aiil fruit wtlile lese than œe meter tall.

The underground parts of many savanna plante are similar and

are coDIDon to cormophytes and phreatophytes having xylopcdia, grubs 1 lignotubers aDi burls.

T:hese pecuUar un:ierground parts may be rel.ated to different functions. The term gru.b refers to the brusby grovth of the aerial parts due to repeated burning but their root systems continue to incresse in size. (Curtis, 1959) • Where the root is loDg and reaches the phreatic water or va ter-table 1 the plant is called a pàreatophyte (Jieinzer 1 1927) vhere the root bas a swelling near 1 ts origin called a xylopodium (Ravitscher, 1948) or Lignotuber (Carter,

1929), or where the underground parts are larger than the aerial parts, auch plants are called cormophytes. Pbreatophytes were the subject of a symposium (Robinson, 1952) reporting a vain attempt at 40 tbelr eradication by means ot tlaœ throwers, to ma1nt.a1n the va ter tables.

Lignotubers are common in tbe Myrtaceae am Proteaceae and act as a regenerative organ atter the tops are burned {Beadle, 194o).

Xy1opcxlia are nearl,y synonymoua vith Lignotubera. They may provide a store of starcb or water (Walter, 1939, Kllllan and Le., 1956) wblcb

is used durlng the dry season. The term bur1 ls syno:nymous vith grub and is used by Metcalte and Cbalk ( 1950) in re terence to the Bricaceae.

The roots ot savama plants have been s tudied by Rachid ( 194 7) ,

Bawitscbar et al (1943), Rawitscher and Rachid (1946), Rizzini am Beringer (1961, 1962) am Donselaar (196-). 41

Tbe fauna of the savama is strongly influenced by seasonal

ebange. In the dry season the scarcity of water and the intense

radiation exchange at grO'Ul» level (as much as 45° c. in 24 hours)

have prod.uced either erosion or heliophillsm. In sp1te of the

presence of an ememic flora there are but :t'ev etldem1c animais in

the savanna (Ba tes, 1948) • Many dry season animals live in deep

burrows 1 e .g. Lizards an:l Iguana (Ve.nzolini1 1963). Most birds

nest at the eoi of the wet season. Fish-eating birds (M;rcteria1 etc.) am Seivingbirds (Cairina, Demrocygœ, Nettion) :find abund.ant

food in the shrinking lakes am creeks • Fruit and seed ea ting birds

{Conurus, Eupsychortyx) :t'ind most :food at the beginning of the dry

season. The predators (Falco, Tachytriochis, Milvago) are most easily seen near savama f1res ei ther perched on nearby trees or

struggling Vith a writhing snalœ on the ground. Vezey-Fitzgerald

(1934) lista over 100 birds trom the Rupunun1.

Pish are most easily caugbt during the dry season vhen the va ter volume is small. The world' s ls.rgest :f"reshwater :fish, the

Arapaima, occurs near Karanambo. The more conspicuous f'ishes are

Pirai (Piranha), Electric eel, Sting Bay, leukananni (McConnell, 1958).

The re are many insecte ani Blackflles and oosqui tos which may distress in the wet season. Malaria occurs in the Rupununi savanna.

The t'irefiy 1 Isneus fatuus 1s common in wet places.

Some mammals are associated wi th term.itaria. The anteeter

(M.y::J.pecopbagus) eats termites, the Armadillo gives birth to ber qUI!Idruplets in a burrov beneath the termitaria, faxes (Dubicyon) 1 42 Lizards and snalœs also she1ter there.

Eartllworms are not coumon in the Rupununi although they bave been blamed for hogwa11ows at Ebini (Stark, 1903 personal. communication),

Aripo (Richardson, 1963) and in Suriname (Voorde, 1957). Apparentl.y

the worm bu.ilds a llummock to keep i tae1f above va ter. These huœocks

are the start of micro-relief format-ion and are called "Kawfoetoea 11

1n Suriname •

Unintelligent hunting b.a.a reduced the deer (Carr:tcus fBVM&M) population. Brock, {1963) presents acute persoœ1 observation on tb.e ma111D8lian fauna. Some of the sma11 (3 feet higb) deer have taken to

the forest so the re is very 11ttle grazing in the Rupununi otber ths.n

that by cattle. Loxton (1958) reports a stock carryin.g ratio at about 70 acres per head ot cattle at present. This 1s probabl.y lees

tb.an the carrying capacity 1 consequently overgrazing is not cOillllœ.

Legg and Rewson (1962) report on beef production at Ebini and Rupununi.

Soil fertilization is being successfully tried at Ebin1. The coat of transport rer.ders this impracdcal for the Rupununi. The relation of man and the savanna bas been studied in detail by 'W'addell (1963). 43

A curirus feature of tbe savanna is the spo:redic occurrence

of termites. The insecte tbemselves are incœspicuous but their

termitaria are Obtrusive. Tbese termitaria vary in size from a

tev inches higb and vide to spires as ta11 as 15 feet vith a diameter

at the ground of l2 feet. In parts of Africa the term1 taria cover

up to 3fil., of the ground (Kellogg and. Daval, 1949). They are so

important in places that tbe savama is eponymously named Termiten­

savannen. (Troll, 1936, 1950). They are found me.inly in savanna

(Beard, 1953), but Harris (1961) says that the Rupununi species only

build on ~he ground due ~othe absence of ~rees.

The termi t..es occurring in the Rupununi and Ebini savannas

are Ifasu t.i :.ermes ephra :;ae { Sands, 1964 personal communication).

They drown tbeir enemies in spittle. These termites eat wood but

an~ digest. a proportion of the cellulose (Harris, 1961). The1r

termi tar1a are bu11t with a thick ruter layer of earthy ma~ter,

probab~ digested, am cemented vith excreted llgnin. The inner part has more excreted lignin and cellulose ani l.ess soil. Myers

(1936) reports 5itemes exceMens am Syptermes snyderl :from the

Rupununi, both of which are in t.be same family as Nasut17-ermes.

Adams on ( 194o) bas reported Nasuti ternes ephratae from the Trinidad savamas of Artpo, Mausica atd Piarco. Ramia (1959a) found N. conû.p.er in the Llanos. Martyn (1:;131) reports the mounds at Ebini as nesta of the Coushi ant (~ sp), which is disputed by Myers

( 1936a) • Aband.oned term:l:t..aria may be recoloniaed by o:her termites orants. 44 Te%11l1 tes have a profou.rd ef'fect on the so11. Grassé ( 1950)

reports tbat termites may collect clay :t'rom 5 meters deep and 300

meters around. Wb ile Bye ( 1955) reports clay extraction up to l2

œters in A1'rica. Besse (1955) found tbe.t termitar1a were formed.

solely of sub-soil1 the chemical properties of which remain unaltered.

llold~y ( 1933) found about 3Cf/, Carbon and 0. ~ Hi trogen in the mounds

but does not ind.!cate the composition of the surrounding soil for com­

parison. The composition of termitaria and soils has been compared

also by Joachim am Kandiah (194o) and Murrows (1938). Wbere the

sub-soil is better than the top-soi1, then the termitaria are ricber

tban the top-soil am conversely. 'lhe same app11es to the 'acculi!Ulation

of lime-rich water (Besse, 1955). The use of broken up termitaria for

:fertillzer (Taylor, 1942) will œly be beneficia! where the top-soil

is poorer than the sub-so!l. Burnett (1948) reports the custom of

piling wocd on termiteria to obtain manure in the St:dan. Troll (1936)

8Dl Barris (1961) report cult!vat ion of crops, restricted to mouDis.

Erhart (1951) attributes the occurrence of a primary laterite

horizon to the quaternary foss111zation of termitaria am calls the

latter .. Cuirasses termitiques''. 'lhis bas been refuted by Gr! ffi th

(in Harris, 1961) and Sillans ( 1959) •

The activities of termites are reflected in the vegetation.

Drwmnom (1.886) draws the ana1ogy between the preservation of the

so11 :tertill ty by termites in the trop! cs wi th the earthwoms in

teœperate countries. Jiye (1955) estimates that half a ton of earth

is deposi ted on the so!1 annually in parts of A:f'r!ca. In the

Rupununi, active mounds are bare; ab&.Dloned am eroded mou.zds support a distinct vegetation (My-ers, 1936a). Many termi tar!a have a tree 45 growing through them but the IDOUDd was probably bui1t around tbe tree, not the converse. 'l'his is tbe opposite of Oinonen 's (1961) observations on .Antbills. Troll ( 1936) explains tbe occurrence of nœ.-savanna vegetation cm termitaria as being relict vegetation reœining before the savanna exteDded into the forest. Martyn ( 1931) reports a com· plete succession frœ saVIUUl8 to forest on termitaria at lbini. Troll's te mi te bush islands are moze related to Gallery forest than savanna species • However in the RupunWli no nœ-savanna plants wre obser.red on termi ta ria.

Little ecological vork has been done on termites. Tbe vegetation of termitaria has been studied by Murravs (1938) and

Wild ( 1952) • Sams ( 1961) bas started quantitative work on tbe spatial distribution of termitaria and has fOUDi sorue relation to habitat. Knoppe • s ecological study in Suriname is eagerl;y awai ted.

The significance of termites 1s that they partiall;y œgate tbree of tbe causal factors of savamas. Tbe termites 1 by tbeir burraving1 improve drainage ard aeration, thus reducing waterlogging.

Termites may increase the wtrient level of the soil, and. may reduce

:f'ires. Termite burrovs may provide a passage through a hard pan for the roots of phreatopb.ytes . MB'l'BODS

PRIPA.R.N.riOB:

Because tœ 1\lpmi saV&Dl'l& is ao Nmote a:r:ld relatively

inaccessible, a prellmiœey auney to obtaln an illp1'eBBion of t.be

nature of tlle saft.DJJ&, vas impossible. In apite of tbe possible

bene ti t of lengtlly &Id detail.ed preparatioas tbese vere 11Dpract1cal

oving to the 1mm1œnce of the dry seasozh

A card 1Diez of plant species vas ccmpiled t.roaa tbe llmited

li te rature available • '1'o this vas added the ~s ot tbe taad.lies

of tœ genera noted. 1be t'ev descriptions avaUable 1 ewn it only

of tbe gross physiognomy 1 and. any eco1og1ca1 information vere added

to tbe cm index. Tbere 1s no lilnua1 or Flora of tbe region.

Bowewr, soœ relevant papers vere obtaiœd, namely Graham (1933) 1

Hitchcock (1922, 1936) 1 Hitchcock aD1 Chase (1917) and Fansbave

( 1952). With the aid of tbese am Willls (1960) 1 1D08t specimens vere 1dentit1ed to tamily. A surprising number of tœse specimens vere assigued to their correct genera by elimination trGm the species l.i.st.

A species list tom tor recording data was comp1led a.u1 mimeograpbed t'rom the card iDdex. St&Dl Record am Tree Data torms vere drawn up. n:.e species list was enlarged from tlle tiliwrsity of the West IDUes berbarium in Trin1dad and some ot tbe species vere learned t:raD the berbarium. n:&LD llmlODS:

Upon arrival in tbe savanna 1 t vas decided to deal ool.y w1 th 47 savaxma 8.Dd to omit :f'rom tbe present study the collecting of rom quantitative information ; It&' Swamps, Bush Isl...aals, Ripe.rian 1 Forest am permanent Marahes. Tbe study or the se areas vas un·

feasible, beillg .fioristically' and .p.bysically too dif'ticult for .. the tilDe available. Furtber, i t wu cons:l.d.ered that time spent

on these possibzy extreae types weuld be most usetu.lzy spent cm

otber types, covering appreciable areas • Tbe nathods used had

to have an accuracy reconcilable with speed, had to be suitable

tor use by a single worlœr; nei ther œed nor assume any prior

knovledge ot the v.getation 8Dd tinally must be quantitative, :re­

peatable and comparable.

gLIC'l'IQlf OJ' STA!IlS:

A nwlber ot criteria vere establisbed to reduce any sub·

jective selection of stands althoqgh the vegetation vas ao unfallliliar

tbat tbere vu 1:10 opportuni ty through lengtby observation or experience 1

to develop preconceived. 1deas about it. The stand bad to be of su:t­ ficient size (about 10 acres) to eliminate from the ares sampled any edge effects • Pu.rtber, the stand bad. to be of uniform au:1 homogeDQUS

pbysiognaay vith respect to topograpby, cowr, tree ani herb distribution, dis turbance (tire, grazillg etc • ) ani appa.rent mois ture cODdi tions •

Stama vere selected to attempt to include aU aspects of tbe vegetational COII8m1 ty. It was hoped that the foree ot any obJect­ ions raised by the method ot stao.'l selectien would be 1111ted by the large number of stand.s used. In practice 1 the vegetatian was fCIUDi to be so un1:torm in large areas that bomogeœity vas not a problem. 48 OOLLICI'IOI' OF DATA t

Data vere obtained both co tbe cbaracteristics of tbe

environment 8.IId on tbe composition of' tb.e vegetation. J'irstly, putative stands vere surveyed to determine their validity. During this initial survey all plant species vere recorded on a presence data form. This presence list vas maintained by using a serially numbered field. name for, and by the collection of' any unlmovn or vegetative material as VCIUchers. The plants were named at the

Smithsonian Institutiœ. In addition, tbe topograpby', slope (direction and degree)' nature am extent ot any disturbance, mo4.sture collditions 1 physiognomy and otber envira:uaental. factors were noted on a Stand Data 'bheet. The se observations vere continued. during tbe rest of tbe study for each sta.J.Xl.

Sampling: Sampllng met.bods vere used to save time wi th tbe hope that tbe sample would represent the whole. To this em a series of

Points vere established illdepellQ.ently of tbe vegetation, along a straigbt liœ (by sigbting on a distant object) • A point vas estab· llshed on the ground every 50 paces. Points ..rere thus rani()lllly distr1buted througbout the stao:l. When the edge of the stand was approaohad 1 anotber compe.sa direction vas chosen; 20 points were sampled. This vas cons1dered to be a more practical method of establlshiiJg points than the use of' r&Dd0111 tables or an actual string grid. At each point a presence list of' species occuring in a meter square quadrat we.a made. Frequency vas subaequently derived trom tbe data so recorded in the 20 quadrats per staid.

R>IIT-Qt~AR'l.!at amon:

The tree am 'àapllDg oomposi tion data were obtained. by tbe 49 Point-Quarter metbcd or Cottam and Curtis (1956), Curtis and Cottam

(1962), Curtis (1959). Because or the depauperate nature ot tbe

woody vegetation a tree vas detined as any vocdy stem of 12 sq. in.

basal ares. Basal area was used 1n pntterenae to breast beight due

to tbe lov, deliquescent branching or the trees. A sapling was talœn

to include all woody stems over l in. diameter at base but lesa tban

12 in. basal area (l-4 in. diameter). At eaeb point the distance to

tbe nearest sapllng and tree, their species and tbe basal area or tbe

tree vas noted in each or the tour quadrants •

At points lO and 20, the pli or tbe soi1 vas measured wi th

a Trueg Field pB kit. Tbe soil to about l2 incbes deptb vas descr:Lbed

briefly am a sample collected.

The water retaining capacities of the soil samples vere determiœd in the laboratory. A sample of each soU vas lightly grouni sieved. to 2 m. and put in a pre-weisiled clean Hilgard eup 1 · of Vith filter paper am lef't for 8 hours dipping in a tray/distilled va ter to beccme vetted. Each sample was blotted tree or excess va ter, veigbed 1 beated at lD50c. f'or 72 bours aDd reveigbed. wet wt ... dry vt. x 100 W.R.C. ~ dry wt. • earpty wt.

The soU.• vere analysed tor acidity1 Calcium, Potassium

Magnesium and Phosphorus (as P ) and organic matter by standard 2o5 procedures at the Que bec Provincial SoUs Laboratory, Ste. Anne de la Pocatiere.

Two samples or termitaria vere talœn (Rupununi and Ebini) and aD&lysed in tbe same way as the soU samples. 50 TREA'DI!NT OF DATA:

A total of 50 stands vere sampled 1 45 in the Rupununi, 2 at Ebini am 3 in T.rinidad. Field data vere ta'bulated and ordered onto

stand auomary sheets. The f'ollow1ng vere calcula ted:

Frequency = ;, quadrats occupied 'by a given herb or shrub. (J: value per apecies per staal).

Actual Frequency (A.F.) . no. of points of occurrence x 100 Total points sampled

Relative Frequency (R.F.) _ no. occurreases of a SR· x lOO Total no. occurrences all spp.

Relative Density(R.De.) == n9. individuals ot l BR· x lOO Total no. indiv:iduals of all spp.

. R.F.::; Preque&' ot 1 species x lOO Total quency of all spp.

Relative Domiœnce (R.Do) :::; Total basal area of l sp. x 100 Total basal areas ot all spp.

Importance Value (I.V.I.) :::: R.F. t- R.De. t- R.Do.

Absolute Density: a) Sum of the distances from 29int to tree .. Average Total no. trees taken - Distance

Total Deœity (no. trees per acre)

c) no. trees pér acre x B.De.::: Absolute Density for each sp. per acre.

Daainance : a) swa of basal areas .-::. Average 'basal area per tre no. trees

b) Average basal ares per tree x no. trees per acre

::- Basal ares per acre ~ Dominance •

F:requenoy vas calaulatad for herba ao:l shru'bs, the otber values vere calculated far trees; Dominance vas calculated tor tel'lli taria. 51

TERMITES:

Termitaria vere found to be physiognomically and numerically as important as 'trees and so 1 t vas decided to treat tb.em in the same vay as wre the trees. The longest basal d.iameter and the distance to tb.e neares t mound in eacb quadrant was measured. 'lhe assumption that the termi taris vere circular at base was considered suff1ciently accurate for the present study. Sands ( 1961) measured. the longest and shortest basal diameters. l'Urther the height was assumed to be proportional to the basal diameter am the termites were considered to be of tbe asme species in eacb termitarium. Tbe termitaria vere not examined to ascertain if they were active or inactive as it vas assumed tbat long inactive termitaria are eroded awy. Two samples of the termites vere collected and vere identified at the Termite

Research Unit of the British Museum.

MATRIX:

Finally the Matrix method on the phytosociological cbaracteristics of species, Bray and Curtis (1957), Maycock (1957),

Maycock am Curtis ( 1960) , Curtis ( 1959), vas a pp lied.

This ordination tecbm.que can be .perfbrmed using any quantitative values vbere a large number of stams is involved. In this case only berbs were considered. A list of herba was compiled using only the taxonomie entities wh1ch at the time were lmown to be val1d. At the tiœ this method vas tiret applied the speciœœ bad not been identified, so i t was necessary to use 11 fièld names" • If tbere vas any doubt as to 52 tbe validity of an enti ty, am surel.y rœ.ny collections veœ tbe same.,

the entity vas not used in the ordinatioo. This reduced tbe herb

liat to Eb apecies which vere prepared. on a mimeosrapbed matrix data sbeet 1 to t'ac illtate the comperisœs required. 'l'be quantitative valuesused in this study were trequency am

presence. Presence onl.y tor a species in a statd vas scored as oœ.

Tbese "Actual Prequency" values were entered into the Mat rix Data sheet, ODe aheet for each of the 50 stams uaed. To talœ 1nto accOWlt the tact that soœ speeies are more comon than otbe?, each frequency value vas expressed as a pereentage of tbe maxiiiiUIII f':requeDcy value tor that spec1es Vh1cb occurred 1n aU 50 stands. This produced a"modi:N.ed trequency"value. Because the total nulliber of species in a stand could vary greatly tbis "modified fl"equency" vas expressed as a percentage of tbe sum of the "ma:lified trequencies" in each st&D:l. This proiuced the "score val.Ù.e" which totalled lOO tor each s~.

The score values for each staiJd were comparad w1 tb the score values tor every otber stand in the following vay. Two stands vere compareeS. at a time, and wberever tbere vere species in coamoo, tbe lower score values of the comparisons were summed • This procedure was repeated cœparing each stam vith every other stam. This produced a nuaber 1 called. "Matrix value" tbat vas ente red into the

œtrix square of 50 stams compared vith 50 stams. T:be matrix values may range tram zero which would imice.te tbat the:re are Do species iD co11111on between tvo stands 1 to lOO wbicb vould 1D:l1cate that the stams are identical in be rb compost tion. This range of numbers is a quantitative expression of the wgetatiœal similarity betveen stands. 53 Tbe me.trix values 1n each rov of' the matrix added to pradu.ce a matrix sum for each stalld. Di.fterent axes, that 1B,

ditterent st&l:ld œ:ders, can be obtaiœd trca this table. Oœ end

ot the tiret axis 1s the stazd. with tbe lowest matrix sum value -

staal 'a' • Tbe otber end of' this axis, stand 'b' , is tbe stand

baving tbe lowest matrix value with staD:l 'a', i.e. the stand ID08t

unlilœ it. Tbe other 48 stand.s may then be arrane.ed 1n otder

between tbese two eDda by tbe tollowing procedure based on tbe ir

quantitative a1m1larit1es to the end stands. All ot the matrix

wlues tor stalld 'b' vere su.btracted trom the h1gbest matrix value

occurring in cme of tbe tvo a taDla • Tbe resulting valœs vere

added to the matrix valwls for the same stands 1n 'a' and the sum

halved to give an average value for each staM. The stands vere

then arranged according to this ord.er. Stan.is 'a' and 'b' will be

at oppos1 te er:ds • This is the first or primary' axis •

20 Axis: 'lhe lovest value in the 10 axis vas subtracted from tbe

higbest, am the resu.lt halved. This is the mid-point of the 10

axis. From an examiœtion or the 10 to 15 staDis which lie± 5

from the mid point the one with tbe 1owest matrix value (2° b)

witb that stand (2° a) vas selected. Tbese two stands are 1;be most different of the st&D1s near the mid point of the l 0 axis am are the

end.s of tbe 20 axis • The ~ 0 axis 1s tben completed in the same vay as

tbe 1°. From the 1° and 20 axes, the etama may be arraD8I"d spatially

1n 2 d1JDens1ons. This distance between st&llds 1s proport1onal to

tbe1r d!i11a1l.arity. The pattern of ataJiis vas plotted on grapb paper. x ;., 54 This spatial distribution ot atallis ia used as a base upon wbich aay

be plotted any of tbe quautitative data collected.

3° Axis: Tbeoretica~, there is no llmit to tbe muaber ot axes

wbich rœy be used. Rowever 1 in practioe three axes are the 1108t tbat

can be easil.y visualised aDl compreheDd.ed. The thi:rd axis vas obtained

in a similar vay to the 2° axis. The mid point vas obtained trom a

central gl"Qlp of steJ:lds on the intersection of the mid points of the

1° and 2° axes corabi.Ded.

The three axes vere plotted on graph paper in 3 dimensiCilS (p.55) "" (lo x 2°, 2° x 3°, and 1° x 3°). A 3-diœnsiœal model wâs also cons tructed. A base board represented 1° x 2°, into wbich vere nail.ed

50 vires {l per stand) 1 and the length of vire represented the 3° axis tor each. stam. It vas easier aDd. lesa laborious to urrlerstand

relationships on the madel than on tbe 3 graphs necessary for the

plotti.ng of a given quantitative character. Data were plotted on

the madel by means of colom plastic beads plugged an each vire, the

colora repreeenting different values 1 e • g. Red tor high, Pink tor medium, etc. Alternatively the colored beads could be used for mere presence or absence, or different colors could represent different enti ties and ecol.ogical factor magni tmes of particular s tauis • 0 0 ().1 0 55 ...... ·. -0 x l') Ut 0 0

0 (/) -1 ::0 m c 0 -1 o· ().1 0 0 z .. . . 0 . "'Tl

., (/) -1 x )> ().1 0 z 0 (,JI (/) 0 :r....- CD CD

Q. 3' CD :J en 0 0-· :J 0 (J.I en 0

...... 1\') : . 0 ... . x ~

01 0 56 MAT RIX

S\HlARY 011' PR>CIDUBB

1. SCOIΠVALUB

a) Prequenciea (:r) ot 8o app. in 50 stands

b) Modit1ecl ll'requency (~ max. J'. tor each spp.)

c) Score value (S.V.) (~ per spp. ot Sum ot Mat. 7.

2. MA'l'RIX

a) Compare S. V. ot 2 stal!ds

b) Su.m the lovest ot co-incident S.V.

c} Repeat tor all species

d) Obtain œe Matrix value tor each staul

Comparison (50 x 50.::.. 2500 - unity comparisons (:50) 2500

2500·50·" 2450)

e) Sum the Matrix values (M.V.) for each atam:...· MBtrix Sum

(M.S.) {one per stam}

3· 1° AXIS a) Lovest M.V. stam::: staul 'a'

b) Stand b :::. stand w1th lowest M.v. in st&Dd 'a'

c) Subtract all M.V. ot 'b' :trom the higbest M.v. f'rom •a• or 'b' 4) .Add tbese to M.V. ot 'a'

e) Divide by 2

t) Arrange sta.Dls betveen 'a' and 'b' 57

a) Subtract lowest value in 1° axis trom higbest b) Divide by 2 Mid point of 1° axis c) Stand 2° a stan:l with 1owest M.S. in central group cl) Stand 2° b stam witb. l.owst M.v. 1n stand 2° a e) Repeat trom 3 b above. RISUL'l'S

Eaeh species ot plant ean tolerate certain cCIIlditi,œs onl.y 1 aDl the extent ot growth v1 thin the range ot these comi tions is cal.l.ed

the ampli tude ot tolerance or just ampli tude. Bach species has i ts own discrete aui characteristic amplitude in the almost in- fini te range of' the environment 1 and no tvo taxa have exactly the same amplitudes ot tolerance. The 11 environment" is the resultant of all the conditions which may affect a plant and may be divided purely for convenience 1nto mutuall.y interacting 11 factors .. auch as water cOD:litions 1 competition, and graz1ng etc. Althougb no factor acta iDdeper.dently it is clear that species dif':ter in sena1tivity of behaviour to the various factors • It is not easy to ascertain which factor or group of factors is the most 1nfiuential in restr1ct1ng an ampli tude, for appe:r·ent correlation v1 th a factor may in tact be correlation vith an interacting factor. As seme tactors are more easily measured than others, these teDd ·to be cited as the most iaportant tac tors • 'l'bus, when drainage is cited as being important

1n thil stmy 1t œy vell be that :f'requ.ency -or intensity of buming vhich is related to drainage 1s actually more important altbougb lesa ameJl&ble to meuurement aD1 reaolutiœ. · 59

TRBATMB!fl' OF DATA

Quant1tetive data were thus obteined for the environmental

and the vegetat1onal teatures of each of the 50 stands. A prime

aim in pbytœociological analysis is to relate the two. This may

be tried by arranging the samples ( stania) wbere the date were

collected, into an arder, which provides an opportunity to relate

vegetation and environment. 'rhere are many me thois o:r arranging

stands; aU have their meri ts am the ir supporte~s • The methoi

used may be basically envirODJ:Dentel, such as by moisture conditions;

by floristics such as used in plant geographieal studies ot range &Di

aff1n1ty; by imieator speeies, by using the physioguomic dominants or

life rorms in the vegetetion, or any other crite rion or measure wbieh

ean be applled to all the sta.JXls. The eeology, that is, the relations o:r the various aspects ot the environment, ot the vegetation, of the

tauns alli their mutual interactions, may then be elucidated by cœparing

ani interpreting data plotted on such ordinates.

A briet acecunt of the variws attempts at ordination is included

to indicate what was doœ, in chronological order. The first metbod

used to obtain an order ot stands was moisture candition. (Bray, 1960,

Mayeoek and Curtis 1 1960 1 Curtis, 1959. et al.) • In auch a metho:l each

stam must be assigned to a moisture class; however, lack of experience

in the present study rendered this impossible.

'l'he second attempt at stand ordering was environmental and f'lor1stic • Of' the 50 sta:Dds 1 3 were une quivocally wet, and 4 were unequivocally dry. From these 7 stands wa.s compiled a list ot wbat was boped to be 12 wet ir.ldieator berbs am 12 dry imieator herba •

A weigb.ting procedure, (Bray am Curtis, 1957) was then applled on 60 tbe presence in each stam of tbe 1ndicatw apecies, acoring 1 tor

each apecies 1Ddicating wet conditims and 0 tor thoae indicative of dry sites. The stemds vere then ol'dered according to the D (dry)/W (vet)

tractiœ or ratio. Once again this order was not canclusive.

This D /W tractiœ vaa then mcxlitied to aclmowledge that sa. stams had more species 1n the list than others. ID general wet 2 bad IDG1"e than dry. The formula (D,IW) (DtW), (Patil, 1964) wu used.

This was a slight improvement over p:revicus attempts am led to sCille

fairly interesting gra'Pbs, but was alao not particularly cœclusiw.

This fonn.tla could probably be impl"Vftd by tbe incerporatiœ

ot mere data, as long as the trema of the data are lmewn. Als o,

it 1s ditticult to assign a weigb.t to the factors used. In the 2 formula (D /W) (D+W), tbe number of species 1n the list ia arbi trarily made half as 1ntluential as the D /W ratio.

The logieal cœclusion of this methCid vœld be the deri vat ion of cumberseme formulae tald.ng into accaunt all tactons meaaurecl. An easier vay of achieving a similar goal 1s to exp&Dd the origiDal list of 24 species to as many speciea as possible 1 am to use all of the samplea, not just 7. rus becomes similar to Motyka'a coefficient,

(vide Curtis, 1959) 2W /a b • ( vbere W= sum of lover trequencies in coliiDCill, a = :frequency in stand a of the cOIIIlDOll species, aDd b = same in stam b.) This 1s, in tact, the basis of the Matrix. 61 PRJ:VALBNT S P 1 C I E S (vide Curtis 1959)

~ SPICDS 'fo PRES. AV. tj, MAX. 'fo FREQ. P'REQ.

Trachypogon 93 ~ lOO Andropogon spp 93 31 lOO

Fimbristyl1a 93 Qo 90 Aristida setifolia 87 24 65 Galactia schomburgkiana 84 24 90 Paspalum spp 75 11 100 Axonopus chrysites 73 24 95 Borreria verticilla ta 73 18 6o (Cassis) Evolvulus becasanus 67 16 80 Byrsonima verbascifolia 67 13 95 Dicnromena cilia a 64 15 90 Polygala longicaulis 64 13 75 Rhynchospora spp 62 20 100 Mesosetum loliiforroe 60 22 90

Polygala an(JlS tifolia 58 7 50 Casearea sylves :.ris 58 5 30 Phaseolus 56 8 75 Buchnera elongata 56 2 20 Bulbostylia conifera 53 27 85 Stenopnyllus paradoxus 53 22 95 Myr-:.aceae 53 2 15 Ruellia geminiflora 49 12 95 62 PREVALENT SPECIES (Continued)

SPECIES 1> PRES. AV. tf, MAX. j, ~- FREQ.

Pbyllanthus niruri 49 11 6o Scleria ("large") 49 8 55 ( "lax") 49 11 55 Erythroxylon 47 1 10 Crotalaria sp. 44 5 46 Palicourea rigida 44 1 25

' Loranthus sp. 44 11 8o Oxalis schomburgki1 42 4 4o Zornia diphylla 42 6 4o Cissampe1os ova1ifo11a 42 1 10 Cassytha americana 42 6 50

Amas onia s p • 38 2 25 L1pp1a schomburgkiana 38 3 30 Paepalanthus sp. 38 15 100 Stylosan hes sp. 38 6 lOO Tibouchina a.spera 38 9 53 Richardia scabra 33 5 4o Panicum sp. 33 10 85 Eriosema violaceum 31 3 4o

Randia formosa 31 ~ 5 Eragrostis glome rata 29 1 100 Sida sp. 29 2 50 Buchnera palustris 27 5 lOO 63 =-P-=R~E_V;....;.::A•L;:....;:;E:....:N:;.;....;;;T;,...___;;;S....:P:.....::::E~C _:I;;....=.E...;;.S (Con tinued)

SPECIES r{o PRES. AV. r{o MAX. ~ FREQ. FREQ.

Cyperus sp. 27 2 Setaria sp. 27 1 35 Waltheria sp. 27 5 65 64

MAIB TREE SPECIBS DATA

Tree , PRIS. AV. MAX. Bo. Times I.V.I. I.V.I. Leading tree

Curatella americana lOO 51.1 300.0 29 Mïrrtaceae ( "pilosa•) 83 13·3 139.4 2 Plumer1a inodora 75 10.5 103.2 2 Byrsonima crassifolla 70 8.8 ao.o 4 Bowidichia v1rg1llo1des 70 6.3 68.1 Erthroxylon ( "flalcy") 50 2.2 62.3 Roupala sp 30 3·1 104.3 1 Antonia ovata 22 1.7 12 ·5 Genipa amer1cana 17 0.5 13-1 Rœaceae ( "Hirte lloid") ll 0.5 33.1 Tabebeu1a sp ll o.4 22.1 Anarcardium occidentale 6 0.5 JR.o Céreus sp 3 o.1 10.5 111rtella racemosa 3 0.05 3·5 65 Although the initial grouping of stands was subsequently discarded, some of the resulta are presented. The trequencies of the spec1es under considera-

t!on were plotted graphically with t~ horizontal axis representing the 5 moiature segments:dry,dry-mesic,mosic,wet-mesic,& wet. As is usual with this method,the

~ frequency curve iw more sensitive but laas wide ranging than the presence curve. :Exfl!lples ara takon from trees,shrubs,forbs,grasses and sedges.

BEHAVIOUR OF THREE TREE SPECIES

50

OVATA importance value index . . .

11 Il H!RT[l LOID

20

AME RICANA 10

WET stand o rder DRY 86

FREQUENCY - MOISTURE GRAPHS,selected speciea'(Jiax.9QO)

PSIDIUM SCLERIA GYMNOPOGON ECHINOLAENA GUAJAVA c•aaroe, FOLIOSUS INFLEXA

SIDA SP. SCHIECkiA RANDIA SCLER lA ORINOCOENSIS FORMOSA ("la x")

200

0 D DM M WM DM M WM W 67 MA1'BIX

TD llATURI OP '!BE AXIS:

Tbe axes ot the matrix are based 011 tbe phytosociological

characteristics of herba alone. Bence the cloaeness of stands

il.ldioates tbeir degree of phytosoeiological similarity. The phyto·

soeiological characteristics of herbs are influenced by all the factors

of the environnent, including the vegetation. Tb.eir occurrence 1ndicates

a specifie re.ne,e of tolerance for a pe.rticule.r comb1nat1on of factors •

The axes actually are phytosoeiological similarities and differences not environœntal factors. If an enviromnental factor does corre la te

thia means that this factor influences the phytosociological relationsbips

of the plants; tbat is all. Bence the axes do not re present single

factors, iaieed. it would be surprising if they did, but probably a com• bination of several or many factors. Most of the factors of tbe environnent, if tbe environrœnt can be resolved into factors, are inter­ depeo:lant. Wbere two or more factors are in direct proportion, tben \.,._ ....--" they will be components of the same axis. 'l'he axes are not tbougbt to be llnear.

By plotting data for all the factors measured., certain recurring pattel'tls become apparent. In the interpretation of tbese patterns it is assumed tbat tb.e axis on wicb. the gradient of t.be factor occurs, ie the axis of tbat factor. Tb.at is, if a factor is very low near the origin and ranges to very higb in tbe lower rigbt band corner, tben this factor is a compone nt of the first axis • By this method tbe na ture of the parameters was aœlysed. The nature of the second axis was easily discernible as representing soil nutrients. Tbe ~ Nitrogen, Potassium,

Calcium, Magnesium and B. were all low œar the primary axis and became 68 hie;ber along the secolldary axis. (Page 78).

Tbe primary axis vas related to the moisture and drainage

cODditions of the ·staDis. However this 1s not the sole factor as

tbe range of moisture comitions is :from very dry near the orig:in to

very wet at 30° from tbe origin. If a llne 1s drswn through tbe main

range of the factor gradient from the law values to the high values tb.en

the angle 'betveen that liœ am the primary axis is a measure of the

degree to whieb that factor 1s a compoœnt of the primary axis • An

angle of 45° would indicate that the factor conaidered was an equal

component of the pr11!18.ry am seconde.ry axes • It was deduced that

the primary axis is controlled to a minor extent by soil nutrients •

The tertie.ry axis was also a coui>inatiœ of factors. Tbe

greatest correlation was vith water retaining capacity am the pli of

the soil. Thu.s the extremely acidic sw:ds were low on the 3 • axis,

r.tsing sUgbtly to the lesa acidic, very dry sites am end.ing witb

the least acidie, rich, wet sites. Tbere is also sUgbt correlation

vith soil nutrients and moisture. Many ecologiste disagree concerning

the nature of the structure of vegetation. Bot coly is the concept

in dispute but the means by vhich it can by explained., or vhat information

cons ti tutes an expl.anation of vegetational structure 1 varies widely from several extremes. McMillan's anal.ogy (1959) is most appropriate, particu.larly if the environment is considered to be composed of so many :l.nterrelated fa.ators tbat the oversimplltied three para~~~eter

tetrahedrœ becomes a mul.ti-polyhedron or spbere. Bovever several re la tionshipe are demons trated by this method 1 inclu:ling the mutual relations of species and their ecological amplitude; tbe relation of vegetation and its compooents to factors of the environment and 69 the continuous nature of' the vegetation. Me.ycock (1957} explains in ' F detail the interpretation Df matrix data. Wbenever tbe matrix bas

been used previously, the stands have been assigned to a moisture

segment before the calculation of the matrix. Bere 1his situation bas

not been possible due to the complex nature of moisture relationsbips

in the savannas, at least during the pericd of observation. The stands

have been grouped from the model into various mois ture categories. As

usual water is an important influence on the vegetation. In the

Rupununi savarma however 1 the va ter conditions are complex in tb.at dry can mean either absence of water or excessive drainage. Some

stands are inundated in the wet season and support a "wet" vegetation,

but these stands dry out so fast aa:'l so thoroughly tbat they are

ecologically dry. The dry season mt y be the period when the se group-

ings woultl prove meaningful.

One feature of savanna vegetation vhich may be eluc1dated

is the varJing occurrence of trees. If' tree densi ty {No. trees /acre)

is plotted on the model, a partial explanatian is evident. The line

joining stams with the greatest density of trees is nearly parallel

to the 2' axis • This inUcates that there is a very restricted optimum

1 tor trees on the l' axis 1 e.nèl that factors on the 2 axis do not in•

nue nee the trees so strongly. The optimum development potentiel for trees is a discrete amount of moisture bere designated wet-mesic.

There are trees on the pisolltic sites, preeumably due to the absence

of inundation, good drainage am better root aeration (vide Ramia,

1959b). It is perhape bere that the trees need am use the pbreatic va ter more than 1n the wetter sites. Trees are excluded from sites where inundation occurs tor long in spite ot the presence of vater. 70 'nlis is indicated trom the line joining stands vithou.t trees. The staD.\s vitbout treea or vith a very law tree density are tboee wbicb are very dry or vbere inundation is due to poor drainage leading to

.._*ration in the soil ,occurs.

The preceding paragraphe reter to trees in geœral. llowever not aU tree species react to the envirœment in tbe same vay.

Curatella bas ita optimum in wet and the wet-mesic stands, wbereas

Bowdic:hia is low in the wet-mesic but higb in tbe mesic aa:1 pisolithic sites. PJ.umeria also bas its optimum in dryer sites tban Cu.ratella.

Byrsonima cr&ssifolia bas its optimum in the mesic am the dry·mesic

sites but is law in the vet. Of the trees of lesser importance 1 Genip occurs in wet stards, Tabebyia in mesic and Antonia in tbe dry-mesle stands. A aimilar restricted pattern is evident for the shrubs.

T.bese patterns are most meaningtul. The positions of the in the ordination stands/are controlled solely by the characteristics of herb composition.

That the trees sbow patterns wben not used 1n the ma trix indica tes the close relation betveen tbe ground flora and tbe t.rees. This relation- ship probably is not causal, indeed the trees are too scarce to bave an influence on the herba. Tbe trees and the herbs probably react to tbe environment nearly independently of each other1 but t.hey both react to the same or similar aspects of tbe environment.

Most of the herba show correlations on the model as may be expected because t.be matrix vas based on their pbytosociological cbaracteristics. Those t'ev species which do not show correlations are probably invalid taxonoiJd.c entities • This is the case for the genera Paapal,wg and .ApdrOPOejOI.\ and seme otber grasses aai sedges. 71 The tvo species of tbe sedge Bulbostyl1s are vell segregated, B. {Stenophyllus) paradoxa is restricted to the dry ani dry-mesic sites

vhile B. coni:f'era occurs higher on the primary axis • The woody herb

Galactia jussieana is :round in vegetation which segregates high on the

1° axis also. Evolvulus sericeus, Mesosetum loliiforme. Zolpia diphYlle.

and Aristi4a spp. all have the ir optimum expression 1n the dry-mesic

sta~ while Dichromena ciliata, occurs 1n the wet etame.

Rearly all species plotted on the model produced a closed polygonal

pattem, 1n general wi th the higher values towards the center and the

lower values towards the periphery. This atmœpheric distribution is

a three dimensional extension of the two dimensional contour patterns {plane~ and the one dimensional/Gaussian curves. (Vide Bray am Curtis, 1957) •

Imperfections in this basic pattern were probably due to an insumcient

number and distribution of atams being used. All of the patterns were

intersecting yet discrete. Even the soil characteristics are clearly

shawn to overlap. 'l'bus these patterns represent the amplitudes of the

spècies on the three axes used. 'lbere were no apparent discontinu! ties

in these patterns. This indicates that the vegetation sampled formed

a continuously varying series or continuum, and vindicates the method

of ordinating tbe stan:is of this continuum. This shoul.d caution workers

against baa1ng all the1r work on a few areas considered to be "typical". "core11 or "characteristic".

Vegetation is composed of many 1nteracting planta of the same

and di t'ferent species w1 th different ranses 1 amplitudes of tolerance am abundance. As the components dif'ter, then the vegetation differa. In

this study the differences ani similarities in the vegetatiœ itself' are used to arrange the different components of the vegetation into an orderly e- 72 spatial pattern. When the pattern emerges it 1s obviaus th&t pbysical f'actore of the environment affect the vegetation to a great extent.

Tbe resulting matrix is entirely objective as it is based on quantitative data 1'rom ~ny sites for many components species, vbich are "the sole

11 determinera ot the resulting patterns obtained • {lfaycock, 1957).

Tbat similarly meaningful patterns were obtained trom species {mainl.y

the trees) tbat vere not used 1n the construction or the matr1x1 is :f\lrtber evidence of 1 ts usefulness and valid.ity. ,_

C"")

48

10

49

50

2

3

9 8 6

4 1 7

5

•

1

STAil)

e

2

3

4

5

6

7

8

9

lD

•

MAT

RIX

•

(MATRIX

•

VALUES)

•

•

•

•

1.6.8

49

50

MATRIX

SUM

VALUB

e 74

G) ('0 ::;:, ·. .;;· 0

0 3 0 ·. .... ('0 x ...... N .. 0 0 .. -~ 0 ::;:, Q

01 3: z 0 ::0 -i ::0 l'Tl l'Tl (/J -i ... 0 c::r --~...... ('0 3 , ·. "0 c::r 0 ('0 ..., c ·. Q 0 0 -::;:, 0 UJ x (1) "0 (,)1 . 0 < 0 c (1) ..UJ (,)1 Q,. 3 ('0 ::;:, UJ 0 ::3 Cl)

l> ::s 0 -::;:, 0

...... 0 ':'.... < ·., 0 N ·...... i 0 • .... 'Co ·... ,,,... . 0- x ·. 01 0 Ln ......

JO x 20 1° x 2° 1° x 2° aulbostylis conifero>75f 1 B1rsonimo crossifolio i.v.i. Curotello ame ricana i.v.i. Stenophy llus(Bulbosty lis) porodoxo >80f----

0

r ...... _ _.., oc-.._------1 1 - ...... , r-- 1 1 l' ' , ' .>60 1 / 1 ( 1 r 1 " 1 1 , 1 1 " 1 1 1 ' 1 " 1 : ~~<60 1 1 1 1 , ' 1 "~ 1 1 1 ' 1 1 .....) 1 1 ', 1 1 1 1 "'"" 1 1 ', 1 1 'i 1 ,,1 ' '1

e e ~ BOWDICHIA VIRGILIOJDES MESOSETUM LOLliFORME STAND DESCRIPTION

I0 X 2° l 0 X 2° l0 X 2°

;.,

WET

...... ,., • • vi --············-·······. :... s;··: 1" ' 1 ' s. .... 1 '\.. • • //· / -- 1 ... 1 - 1 : • 1 1 - ....__- • 1 - • • ~... ,, 1 ,~ ~~ a tl =0 ·')M ...... ·w.., 1 , ' - ~---- ·l· ---- • r. ., l''~\.------\...... 0 1; 1 '· ·1 ': 1 ••••

li_~·{·...slope

· /~i stands ')lOf . /·r;J ~ ···... • 1 .... , ·• .. ) 4 4 i.v.i. • 1 1 • )1\ / ·~pisolithic 1, / 1 >\:f.pJ~ A.) stands

e e .· 0 77

.· ·. ··.·.········!······ .· x 0 v (\) U'l 0 0

-f ::u rn 0 rn ~ (/) 0

:'· . \) .. -········: rn ... ::0 ··· ...... -0 · ...... \. x 01 ~--~------~ 0 0 )> C> ::0 (Tl

0 .· .· ~ ·. 0 ·. ·. (\) 0 )( 01 r' 0 ..,..... TERMITES and SOIL NUTRIENTS (K + ca+ Mg) WATER RETAINING CAPACITY •

l0 X 2,0 1° x 3• 2,0X 3°

Termites~ • . . .{ . .. . . · absent present ··...

.: .:

....~.. : _.:·

...:• :.,...... ~· . . .: ::. ~: u-)55°/o ~ .· : \!.•'" -:.

~···41·= ·! -- - ., \ ...... ------, ---­-- ... , '\ --,,,' .... ::: Soil Nutrients ...... 1 • t• \ •' . ·. \ ,' '. ... . · ·...... " / // ">55°/o \ / . \ \ . t 1 ..... ':{r-- > • 1 \ \ ' 1 · >\ ~<.250lbs 1 acre \ \ \, ..... ( { ' ...., 1 \ \ t :.~..·•···· ..-:: . \ \ ;.,...... -'( 35°/c 1 ' \ \ f 0 \ 1 Î '~<35°/o ' ) a.. \ '• , '""', \ \ ,' ',,, \ .

e ., e 79

The presence and absence of termitària were plotted on the

mode.l and were found to be sut:f'iciently sensitive although densi ty

and dominance bad also been calculated. There vas a striking

correlation between termitaria and the stand categories. All of

tbe stailds designated dry-mesic bad termi taria except one statld..

Tbis imicates that termites have a marked preference fœ these sites.

Though not restricted to these sites a few termitaria were found in

closely contiguous categories. to All three axes seem to be relata/the occurrence of termi taria,

as they occurred near the origin of all the axes. To ascertain what

factors bad the most influence on termites, their occurrence was

plotted against soil nutrients. Tfiere• wae an exact aild inverse

relation between soil nutrients ani termites. Wbere the level of

soil nutrients was over 250 pounds/acre, there were no termites what-

ever.

It is interesting to specul.ate whether or not the termites

remove s.oil nutrients and concentra te i t in the mounds or simply avoid

higb nutrient areas or the factors which produce high nutrient levels.

Some factor may reduce termite po~lations on these rich sites.

Tbese results were considered to be a most illum:1nating glimpse

at the structure of the ecosystem. This is thought to be the tiret

Ume that a quantitative relationship bas been demonstrated by this

methcd between the vegetation, the soil, and the fauna. e. 80 PICTURES OF Tfm THRD DDII!fSIOBAL Hl>BL

(Black Do~ Marks Direction of Origin)

l. son. lt.JTRIDTS: (K + Ca + Mg) Low soil uutrients 1 lesa than 75 l:bs/acre (single black bead) are to the lef't; bigb soil nutrients 1 over 200 lbs /acre are elus te red to the rigbt (double llgbt bead.s) • ·

2. OBGAlllC MAftiR '1>: Low values (blaclr. beads lesa than 1.~) are low 011 ~he ~hirel axis and aeparated trom the bigb values (double Ugbt beads over 2j).

WATER REAIIIIlll CAPACM: Low values (aiDgle black beeds) lesa tb&n 3~) are weil separated froiD bigh values (double Ugb~ be8ds over 55j) • Le:rt photograph ahows the separation on tbe secGild &Dl tbe first axes trcm above; the "right photograph shows the distribution relative to the third axis.

4. AVJBAGI COVD ~: Double light beads ~~sent stands Of over 6Cfl> cover, vhile black denotes und.er ~. lfote the s1milari ty in the pattems ot tbese 1'ou.r tactora • The stade towaJda the arigin are· low 1n Butrients 1 Org&Dic Mat~r, Water Betaining Capeoity 8Dd Cowr. Away traD the origin the situat1011 18 Nversed •

Ruell1a pœ;l.p.ifiora: Beada mark the presence ot this species; this itselt is ta1rly reatricted. l"Urtbermare this presence 1a aepanted 1nto lov t:requency (single black bead uo:ler ~) &Dl high trequeucy ( dougle Ugbt bead over :nf,) • 81

6. Dic~a.a c1Uata: Single black beads iDdicate staads wbere tbe qüêncy 1s lesa thaD 3;1J tbe dcuble llght beads irl41eate trequenc1es ot over 3,;,.

Met"'R!um lol!\etw-' Tbe double light bead.a are œ st&Dls whe:re the quenay owr ~. The empty atad.are ot pretence Ollly. The siD.gle black beads mark the absence of tbia grass • Ilote bov restricted is the aœpl.1. tude ot tolerance •

a. Bgrcaima verbuc1tol1a: The black beads lal"k stands ot over 3 trequency and 1n general are tovarda the l.ower rigbt corœr. Tbe un.til.led rcds 1Dd1cate trequenciea lesa t'bat 3(;11,, am the uey beads irldicate abse:ace • This' apecies 18 theretore very vide ranging, poesib~ due to its habit but its optimum expressioa. occurs ill dry sites.

TRDB PBR ACJœ: Black beada represent over 50 trees per acre. This bas a DU'I'OW' optimum œ. the secoo:l axis (bottom edge of photo) as tbe beaà.s are 1n a close Une parallel to the t1rst axis. The grey beads 11ark treeless stand.s arr.i the UDI'IMU'ked roda are stauis w1 th tever t.blm 50 trees per acre • MaDy of the t:ree-less stands are l.ow on the tirst axis •

lO. BYn0!:!.1-fe C~!!it:fa: Black beads over 60 I.V.I., seven centra c te st&rlds. Cyl1J:Jdr1cal be,ada mark treed staDls vitbou.t !• CJ.!8&1tol1a; grey beads mark t:reeless steuts.

u. llam!!i! 1nodora: · Black beads over 60 I.V.I., otben as no. lD. 'l'bis species 1s lov cm the third axis arad tairly narrowly restricted. 82

12. BcNlichia rtnù.lioid.ee: Blaok beads ,over 45 I.V .I. 1 otbers as no. lô. This apecies ie even more restricted tban Plumet*•·

Curatella amer1C!!!!: Black beade cmtr 200 I.V .I. 1 grey cyllDlers leas tb.an 200 I.V.I. Thil ahowa a cloee co­ r.relaticm oc tbe tiret aud third axas. · AU the high values are gro.aped avay t'rem the lower values •

14. ~~!!l!!icana: smll darlt grey cylJ.DI.ers {5) bê ~. large grey cyllDiera ( 4) !Btcs.H oqyu black beeds { 6) Tbe beads here mark. the presence alœe of tbese minor tree species 1 au ot which show discrete, unique yet owrlappi.ng amplitude a.

15· "•qt1teœa tphr!tM: Black beads 1Dd1cate the presence ot termites. O:rey cylillldera soil nutrienta over 250 l:osfacre. Tbe termites ar. all cluatend together w1thœt any DOil•tel'll1te stltDds intenening at all. In lldd.itiœ they are reatricted &l'CWil the dry mesic area. Tbere 1s alao a Dice separati cm 1'rolll tbe higb soil-nutrient sta!lds; the dividing llœ betveen tbe two 1s Yeey distinct.

u;. ( aa no. 151 trœ above • )

,.' 83

SOILS

From an analysis of the soils and their reactions wi th the vegetation, tbree main categories were apparent. The reddish, excessiv ely drained so11s called Rubrozem by Hardy ( 1962) were mainly dry-mesic. These soils are very deep and very po rous • Bardy ( 1962) designates most of the Cerrado soils as belonging to this red-brawn latosol group, althougb this term has been criticised. (vide Mohr and van Baren, 1954). Tlle ether well drained. soils vere the aams. Wh en excessively d:rained on uplaDd sites they were very ligbt in color; if lower down t.be drainage catena they œy becowe brown and darker. The sandy soils were eco1og1cally dry although the 1ow sand a reas were seasonally inundated. T'œ third main type of soil was dark colored and bydromorpbic • Sucb soils occur in low moist sites and were, generally ricber. These hydromorphic soils were not well drained but true planosols were rare in the Rupununi. In the Aripo savanna in Trinidad very distinct pans are present botb of the clayey mottled orange type wi tb concretions and of the .fragipan type. The drainage in these areas was severely impeded but the pan was so shallow that the soils become desiccated in the dry season.

In general, the soils are very deep ao:.l. water is not 1n short supp1y. In the Rupununi the phreatic level w~s usually above 10 meters at its lowest point. Schubert ani Rswitscber (1950) and Schubart (1959)

:round that the phreatic level was highest in the dry season and lovest in the wet season due to a time lag o:f about six months :for infiltration to 20 metera. Rawitscher, l"erri and Rachid (1943) claim that the

Cerrado soils carry three years rainfall in seme places • The foregoing indicates a basic difference in savanna observations.

'l'he Brazilian schoo1 f'ims phreatophytes and no impedence (Jerrt, 1943,

1944, 1955, 1960, 1961, and Coutinho, 1958, and La.bcuriau, 1952,

and Lambert!, 196o, etc.) while in the Llanos a similar vegetation

occurs on planosols where the pan ofters resistan<:e to the roote of

species wb.ich e1sewhere are phreatophytic. (Varesch1, 196o, P'oldats,

1964, Beard, 1953, ani Sauer, 1950). Tbe main cause of -poor tree growth

is lack of root aeration due to floOO.ing (l'ramer, 1956, Micbelmore, 1939).

This appears to be a partial explanation for the denser growtb. of trees

in the Cer:rado than in the Llanos.

The var1ous aoil factors represent easily measurable components

of the plant environment, and in consequence the soil vas analyaeQ. in

some detail. The vegetation is, ot course, in intima te contao•t wi th

the aoil arxl interaction between the tvo can be expected. Correla tiœ.s

csn be demonstrated but it is more difficult to cite cause a:r:d eff'ect

relationships, due to complex eco1ogica1 integration, the inherent nature

of Bco1ogy. The Ecologist meaaures naturally occurring aspects of the

ecosystem such as the }:'hytosociology whicb is a statement of what existe

· and an interpretation of this state as far as possible. Tbere are oo

control experiments and no factors can be kept constant or eliminated

for the course.of the experiment. So adequate analysis of tbe Ecosystem

is both compllcated and exceedingly d1ff1cu1t.

Without exception, every factor {.)f tbe soil t'bat vas measured in

this study was found to correlate to sorne extent witb tbe three d1mensiœal

vegetation mod.el. Tbese correlations 1.1ldicate relationships between tœ

vegetation and its env1ronment. In spi te of the unusually law levels

tor soi1 nutr1ents, even the sœ.ll differences that existed vere fOUDi to 85 have very significant relat1onships vith phytosocio1ogical characteristics as seen in the stand distribution in the rnode1.

The moist, poorly drained sites were relatively high in nutrients, pH, and organic matter. The excessively draiœd san:l.s am red earths si tes vere 1owes t in these factors • Even when soi1 color is plotted on the mode1 the :reds are clustered together apart from the group of dark colored soi1s. The water retaining Cape.city was especially significant on the third axis, and cor.related vell wi th the occurrence am densi ty of trees.

Tl:lese observations support the theories of Alvim ( 1954) and

Arens (1963) that mineral de:ticiency in the soil is important in savannas. In this respect the situation in tbe Rupununi savanna seems to be more like that occurring 1n the Cerrados tban the Llanos.

The soils of the Rupununi (1oxton et aL, 1958, Stark et al, 1959) are s imi lar to those o'f the Cerrado. ( Ranz.ani, 1963, Hardy, 1962, Viana

and Araujo 1 1946, A1vim, 1954, am Araujo, 1952). Arens• Theory of Oligot:rophic Scl.eromorphism (Arens, 1963) bas been questioned by

Labouriau (1963). Alvim and Araujo (1952) maintain that deep roots are more a reeponse to the law mineral level than to obtain more wa ter. Denevan ( 1964) 1ndicates that tree roots caimot rea ch th.e

:f'resh deep soil so that a closed cycle ot nutrients occurs, further impoverishing the surface so11. 86

ENVIROBMBNTA.L CBARACTERISTICS

son.s , TBRMITARIA Av. Av. M:Ln. Max. (of 50) (of 2) Min. Max. pH 4.4 6.2 4.9 4.8 ' 4.7 5·0 pli tampon 6.1 6.9 6.65 6.7 Di t'terence o.1 0.9 0.3 0-35 (t'rom 7 .o)

~Carbon o.u 9·49 1.47 2.75 1.67 2.48 "fo Orga.nic •tter 0.2 l2 .4 2.5 3·6 2.9 4.3

"fo~ o.o2 0.75 0.09 0.27 o.60 0.14 P2o5 lbs /acre 26.0 60.0 35.4 49.5 30.0 69.0 K Il 3·0 148.0 41.6 155·0 90.0 220.0 Ca Il 0 572·0 90·9 697·0 214.0 1180.0 Mg ft 0 136.0 42.1 293·5 87 .. 0 50Q.O Total nutrients 3·0 689.0 172·7 ll45.5 391.0 19()0.0 (K + Ca+ Mg) H·llilli-equivs. 1.0 16.0 3·7 3·2 3·0 3.5

Wa ter Retaining 28.3 186.2 50·7 4o.7 37·2 44.2 Capacity

%co "'11er 20.0 95·0 47·7 ... 87

FLORISTICS

This list of species probabl.y containa most of the species

occurring in the Northern RupunWli Savanna, and was compiled. mainl.y

from tile author • s collection, determined. by the Smi tbsœian Insti tutiœ in Washington; for these the appropriate authorities are given in tbe list. 'lhe other species are from )\yers (1934, 1936a, l936b), McCorkle

(1952), Rutherford (1958) 1 Harrison (1$63) and Graham (1964). The

Ebini plante have been assiduously collected by Harrison (19::>8). The

Aripo Savannas in Trinidad have been recently botanised by Richardson (1963).

The limitatiœs of the list are:

1. Some habitats were not collected thoroughl.y, e.g. Ité Svamp.

2. Bush Island flora is not included.

3 • The vhole of the savama area was not covered wi th equal detail.

4. Some of the flora at the eni of the dry seasœ vas not collected.

5· The determinatiœ.s vere not complete at the time of writing.

For tbese reasons a statistical floristic analyais is not . presented, but a general accow:at of the flora is included. The 11.ora may vell prove to be the best cri terion tor savannas as it is the resultant of all prevtous classifications.

About 100 tamilies of flowering plants are re~sented, or a ,1 third of the total number of plant familles in the world.. About a quarter of these familles are monocotyled.onou.s 1 the larges t family being the Gramineae, followed by the Leguminosae 1 Cy-peraceae and

Rubiaceae, all w1 th more tha.n 25 species. These four tamilles represent 88

~ of tbe flora. Thirty seven familles are z:epresented by a single species. Most of the families are tropical and many are widespread.

• Tbe Martyniaceae am Bixaceae are ~the onl.y s trictl.y South American

t~milies while the Cactaceat aiid Limaptllaceae are !èw Worl4 familles. , 1 The Pedaliaceae is naturall.y restricted to the Old WoNd ani bas probabl.y

been introduced to South America onl.y recentl.y.

The f'loristic at:f'1nit1es of the Rupununi Savanna are ditn.cult

to determine owing to the l.ack of adequate species lista from other

savannas. Available Usts vary more in degree of completeœss than

in flora. As may be expected the contiguous Rio Branco savanna is • very similar to tbe Northern 1\lpununi am even more so to tbe Southern . Rupunun1 savanna. Most species ·in Talœuchi's lista (1960a and b) occur

in the Rupununi. Barrisœ ',s d.etailed list from Eb1n1 ( 1958) reveals

a s trong affinity to the Rupununi but the intermedia te savannas are

more similar to the Suriname savannas (Beyligers, 1963, Dœselaar,

1964, Lanjoœ,(, 1936, 1954) and the Cayenne savannas. (Bou11lene,

1925, 1926, Rue, 1958 ard Boock, 196o). Tbese Intermediate savannas are flor1stica1l.y related to a lesser extent to the Amazonien pockets of

savanna. (Buber, 1900 1 Ihering, 1907, W.tzelburg, 1939, Ducke and Black, 1953, Rue, 1958, am Egler, 196o). Parts of tbe Llanos, are very similar to the Rupununi savazma.

The comparative list of Lasser (1955) is by far the most complete.

'lb.e shorter lists of Blydenstein (1962) at Calabozo, Ramia (1959a) at " Apure and Bueck ( 1961) show s trong s 1milar1 t1es •

The Brazilian Cerrados, although thousands of teet bigher than

the lb.tpununi, show a marlœd similarity to 1 t. The Cerrado flora

(Rawitscher, 1948, Rawitscher and Ferri, 1942, Ravitscber and Rachid, 89 1946, Rizzini am Beringer, 1961, 1962, &Di Perri, 1955) bas many of the Rupunini species but is floristically richer in general. These

important Cerrado plants have not. yet been reported :from tbe Rupunini

savannaa: Dimorphapira sp, Stl"Yphnc!deDii"?D sp., Au:Ura sp., ICielme;mra.

Caryocar sp., Qualea sp. 1 and Salvertia sp. The savamas in Soutbern

Braz1l show rather lees floris tic s1m1lar1 ty to the Rupununi. (lUten,

1963, Alvim, 1952).

The savanuas in Atrica bave some species aai genera in coamon wi th ;tbe Rupununi. Tbe a:f':f'ini ties are llmi ted to eosmopoli tan aJJd pantropical plants. There is a close pbysiognomic and ecol~gical similari ty however. (Cole, 1963). Adjanohoun (1962) gives a full

list :for the Lover Ivory Coast, am Acocks (1953) aoi Bews (1917) for

the Veldt. Chevalier (1931, 1933) and Schnell (1961) compare tbe tloras of South .America and Atrica in detail. Good (1953) presents useful tloristic analyses, and Vester (1940) maps the range of each family.

Chevalier { 1931) llsts 300 &pecies as coiiiDon to the .American am At'rican tropics aDi wich are not weeds. About ~ of the se

~pecies occur in the Rupununi savanna. So far on1y about 20 Pantropical weeds bave been reported from the Rupununi, and about 8 American weeds.

The genera Helicteres, Icl:mantbus, &Dd Symplocos are coiiJIDQQ to the

Rupununi am .Asia, while Cochlospe:rmum sp., Sympbont.a sp., am Passifloa sp. are discontinuous in the tropics. The genera Cassis, lr:rthroxylon,

Byptis , and Turœra are mainly American.

Many savanœ plants have deep and swollen root;s. Rawitscher

{ 1948) mentions ten plante which occur in tbe Rupununi and bave roots tapping tbe water table 1 and calle them phreatophytes • Tbere are 90 probably many more phreatopbytes but roots are not easy to study

so 11 t tle work bas been done on tbem. It wou1d not be surprising "' if all the Ru.pununi shru.bs were event.ual.l..y f'ound to be phrea t.Oph;ytes

to varyi:ng degrees •

Xylopodia are undergrOUlld swellings of root or a tem, which act as a food or water store 8lld which enable the plant to withstax:td

ti:re 8lld drought. Most of' the low woody shrubs have xylopaiia.

Ten plants tran the Rupununi are listel as good examples of xylopocUa.

lphemerals are bere distinguisbed from aquatics in that the former live iD. flood water &Di tbe latter in pcm:'is which rttay, however, be temporary. Some of the types are facultative. Ten epbemerals ' are llsted as examples of which only !-:9Photos&rpHf sp. alld Bacopa spp. are at all ccxnmon.

The flore. consista œinly of perennie.l.s 1 but may be divided. into 6 groups according to their habit. The trees and most of the shrubs are phreatophytes . Most of the woody herba am amall shrubs have xylopod.ia. (Hoehne, 1939). Aquatic plants are not reall;y part of the savanna. Ephemerals canplete the ir short life cycle in flood vater. Moe t of the grasses are perennials. The roots of the

Rupununi grasses penetrate to a depth ot 1.5 to 2.5 meters and are in contact wi th auch a large volwœ of soil that the plant lasts vell in to the dry season be fore the tops be come brown and sere. J'ire may aometimes stiœuls.te the production of succulent sprou.ts before the rains, 111d1cat1ng that seme stored food am water is present late in the dry season.

The final group of plants coapoaed of aœuals is very small i.D:leed. J'urther s t'L'dy may reduce this group by demons tra ting 91 Xy'lopodia in some members • Theae plants use water in the top layer

ot tbe soil and produce seed rap1dly aoi abundantly.

The Plora in general is very poorly known. Two new species

ve~e collected in the course of tbe study: Scbipea and S!cç1!kpis

ani others may be new. The flora is interesting as it is so varied am con tains sœe beautif'Ul atld unusual plants. ~re are many

species of the insee tivorous genus Utr1cular1f.. aui at leas t two ot

Drosera. The tree used to stun tish, Antonia ovata is cOI:IIBOn. 'lb.e uncommon Baemodoraœous plant Schieckia oacurs in tbe wet lowland si-œta.

It is ot"-...en said tbat the savanna vegetation is xerophytic.

Althougb the vegetation bas some xeromorphia teatures (Morretes am

Ferri, 1959) in general savanna envirœment is not truly xeric.

Most of the supposedly xeromorphic teatures appear to sane extent adaptations to fire. Xylopodia, tor example, probably are more pyrophytic in function than arophytic. J!Bny xy1opcd1a p:}.ants have very deep roo'... s supplying the plant wi th adequate water un til the fire. 'l'ben the xylopodium provides nutrients for sprouting.

In addition the trees are not truly xeromorphic • Ferri ( 1944, 1955 1

1960, 1961a, 196l'b) bas shown that cuticular transpiraticm. is very high, and that the transpiration curve 1n general tollows that ot evaporation. Tree leaves are generally large. The grasses are no more xeropbytia in tbe eavannas than in any other œtural grass- l.aD:l. However, the space in the root cortex reported by Rawitscher

( 1948) may be a xerophytia feature, aai Cac ti are certainly xeromorphic altbougb UDCOtiiDOn. Their reots vere not studied. 92

FLORISTIC A, BAL YS I S

LARGEST F.AMILDS :L.AamST GI1ŒRA FAMILY: JO. OF SP.ICIIB GIIIUS: 10. OF SPBCDS

Gramiœae 98 Paapalum 16 Leguminoeae 55 Panicum 14 (Caesalp1nioideae 12) .AM.ropogon 13 ( Mimosoideae 6) Polygala lO (Pap1llaœt8e 37) Cass1a 9 Cyperaceae 39 Cyperus 9 Rub1aceae 26 Briocaulon 8 Compost tae 19 Azonopus 8 Melastomaceae 15 Rbyoohospora 1 Euphorbiaceae 15 Byrsonima 6 Pol.ygalaceae 12 Phyllanthus 6 Aeschynomene é

P.Alfl'B)PICAL WDDS

Abrus precatorius Crotalaria Alternantbera sess1lis DesmodiWD Amaran thus spinosus Emilia sonchitolia Andropogon leucos tachyus Hyptis atrorubens Aeschynomane Leersia bexe.ndra Bidens p11osa Mimosa pudica Borreria Paspalum conjugatum Cassia Polycarpaea corymbosa Cencbrus echinatus Sida cordif'olia Clitoria Sporobolus 1Di1cus Urena lobata

AMBRIC.AN TBOPICAL WBIDS

Boerbaavia coeciœa C1idem1a h1rta Cynodon dactylon Dactyloteuu1m aegyptium Bleusine 1J:Id1ca Jusaiaea peruviana Maranta at'U.Dlinacea Phyllanthus carolinensis PBYSIOGiiOMIC ANALYSIS 93

Anarcardium occiden. tale Curatella amerieana Myrsinaeeae Annonaceae BrytbrOxylco (round leavea) G:lmaceae Antonia ova ta Erythroxylon (red f'laky) Pallcourea rigide. Bowdichia virg111oidea Genipa amerieaD& Paullinia Byrsonima coccolobaefolia Guet ta:Ma sprueeaœ Plumeria inodore. Byrsonima coriacea Cuttiferae Roupala Byrsonima crassifolla Hirtella racemosa Rutaceae Byrso.nima stipulacea llomalium Bosaceae Cactaceae Bumiria balsamifera SapiDia.eeae Coceoloba spp. Humiria fioribuma Tabebeuia Conœrus inecmptus Leguminosae spp. Tu.rneraeeae Combretum Myrtaceae {privet) Verbenaeeae Copai:f'era Mïrrtaceae (pilosa)

COIIIOK TBDS

Curate lla ame ricana Allopbyllus Lippia schomburgkiana Myrtaeeae ( piloaa) Byrsonima verbaseifolia Maek:aerium' Plumeria inodora Caaearea Mimosa Byrsonima erassifolla Coecoloba , Ouratea Bowdiehia vir~ilioides Clibadium surin8tllellse Psidium spp. ErytbrQxylon ( flaky) Davilla aspera Ran:lia formosa Rou pa la Erytbroxylon Rutaceae ( 'lXan.toxylum) Antonia Galactia jussieana Tibouchina aspera Ge Di pa Guazuma Toeoyena Tabebeuia

PBRIATOPKYTIS

Aegiphila Centrosema Anacbaris Anuonaeeae Cissampelos Baeopa Byraonima eoc:eolobae tolla CocblosperDIUIII Butumaceae Byrsanima verbascifolla Galactia jussieana Cabomba Cœnaru.s HeUeteres Calli triche Copaitera Hyptis Hydrocotyle Brytbroxylon Lippi& Lophotoearpus Mackaerium Merremia Mayac& ~ratea Polygala manticola litel.la Palieourea Ruellia Beussia Toc oye na Waltberia

..1 94

§ H ~ &1 11 Q 1 ~ 1 ~ ~ • ~ 1~ ! 0 11 11 1 1 Cil 1~ Anarcardium oceidèntale 0 0 0 0 0

Antonia ovata 0 0 0

Bowdichia virgilloides 0 0 0 0 0 0 0 0

Byraonima crassif'olla 0 0 0 0 0 0 0 0 0 0 0 0 0

coccolobaef'olla 0 0 0 0 0 • 0 0

1 Chrysobalams pellocarpa 0 0 r 0

CCDDarus sp. 0 0 0 0 0 0

Curatella ameri.cana 0 0 0 0 0 0 0 0 0 0 0 0 0

Brythroxylon spp. 0 0 0 0 0 0 0 0 0 0

Genipa amerie&Da 0 0 0 0 0

Ir Hirtella racemosa 0 • 0 0 Humiria tloributda 0 0 0 ' PaUcCUl'e& rigida 0 0 0 0 0 0 • 0 0 0

Plumeria illorlora 1

Raalia sp. 0 " 0 0 0 0

Rcutala sp. 0 0 0 0 0 0 0

Tabebeuia ap. 0 0 0 0 0 0 0 0 95 ' ~ OC'CURRilCB OF OOME RUPUliJIII PLABl'S e ~ ~ i ~ 1 1i • t.)~ ; 1~ ~ 1~· 1~ C'll 11~ AD:lropogœ leucos tachyus 0 0 0 0 0

sell081JUs 0 0 0 0 0 0 0 0 .. Aristida tincta 0 0 0 0 0 0 " •

JI H Axonopus compreasus " .. 0 0 .. 0 0 0

tt Bulbostylls contrera 0 0 0 0 0 .o

Byrsonoma verbascirolia 0 0 0 0 0 0 0 0 0

Casearea sp. 0 0 0 0 0 0 0 0

Cissampelos sp. 0 0 0 0 0

Echinolaena intlexa 0 0 0 0 0

Eugenia sp. 0 0 0 0 0

Galactia sp. 0 0 0 0 0 0 0

Gymnopogon tollosus 0 0 f 0

Lippia sp. 0 0 Paepe.lanthus capill.aceous ' " " Panicum laxum 0 0 ' 0 0 0 0 Paspa.lum carinatum 0 0 0 0 0 0 0 0 " pulchellum 0 0 0 0 0 " " 0

Psidium sp. 0 0 0 0 0 "O 0

Bhynchospore. barbat& 0 0 0 " 0 0 " • 0

Stenophyllus paradox:us 0 0 0

H Trachypogon plllmosus 0 0 ' 0 0 0 t 0 96

Dm, British Guiana, Harrison, 1958 aild Author's collection

SURDIAMI1 DODSelaar1 1964.

LLAIOS 1 Lasser, 1955

CALABOZO 1 Llanos, Blyd.ens te in, 1962. APU1Œ, Llanos, Ramia, 1959.

CBAPPARALBS, Bueck, 1961·

AMAZOBIAB SavaDDU, Bgler 1 1960 • CURitOO I, Riz&ini, 1963•

CIRRADO II, Ferri aDi Coutinho, 1958• s. BRAZIL, Biten, 1963.

TRDJDAD, Ricbardsaa., 1963, aD1 Author•s collection

HOJI)URAS, JobamaessCII1, 1963.

RIO BIIARX) 1 Takeuchi 1 1960 • CIUDAD BOLIVAR am vicinity, Corp. Venez. de Guayana, 1961.

0 Present ' Genua pneent • Nore tban oae otber species present. 97 DISCUSSION

A distinctive feature of the litere.ture o:f' the savall.Das is the

search for a grand theory to explain their eXistence aa1 nature.

Howewr, it seems unreasœable that one cause could be respœsible

for all of the many different types of savanna 1 which tode.y are the

result of a combin&tion of m&Dy' factors. It would seem more appropriate

to study the problems of each savazma separat&J.y. A geœralization may

then be sought wben i t is based on a su:f':f'icient number of s tuclies of

different savam.'IU.

At present, auch genera.J..izations trequently cœf'Uae dit:f'erent

aspects ot the problem. A clear distinction should be made between

Predispoeing Factors, Causal factors, Resulttng Factors, &Di Ma1ntaining

Factors • Bach type of factor is susceptible tO different types ot

s tudy. This study is mainJ.y concerned w1 th main taining factors, tha t

is, the factors operating in a particular savanna at present.

Tbe main theories concerning saV&l'Dlas tall into three categories:

Climatic, Biotic and Edapbic.

Cl,imat1ç:

Formerly there was thougbt to be a' savanna elima te. This is no langer considered to be so (Beard, 1949) as savanœs have been touni

to occur 1n many different climates. In geœral, a dryer elima te may be a Predisposi:a.g Factor for savanna formatiœ in some areas. Dr-.ter microelimates may be a Resulting &Di Maintaining factor due to tbe removal ot forest.. Gross change in climate is considered with the ldaphic theories • Water def'1ciency is not thougbt to occur in the savannas studied by Bawitscber, {1~, 19481 l950a), Rawitscher ani 98

Ferri, 191t2, Rawitseher, Ferri and. Rachid, 1943, and Varesehi 1 1960,

ete.) • Certainly exeess or paue1 ty of wa ter does not 11m1 t the

savama ground layer as mueh as low light intensi ty limi ts t.he forest

ground layer. Tbe effeet ot the alternation of wet and dry seasons

has been amplified in savannas and so ma'y be a Maintaining and Re­

sult1ng effeet. This does not apply to the savamas of the Amazon basin. Wind is a resulting factor as 1t dries the topsoil after

the rains 1 inerease3 in treeless areas and increases the effec ~ of tire.

Biot1c:

Man is an important Causal Factor in savanna formation due .o

his cutting and burning of forest. Jl'ire is probably more of a Re­ sulting Factor than a Causal Factor of savanna, but it is definitely an important Maintaining ll'actor for many such areas of savanna. Oœ distinctive etfect of tire is the abrupt nature of the savanna-forest boundary. It may also eontract any intermediate areas between the two. Grazing am trampllng may be a Maintaining Factor in some savannas • Oecasional overgrazing vas observed near ga tes in the

Rio Branco Savanna, vith a very depauperate vagetation of wedy plants especially ~ spp. lliaph1c:

Impeded drainage bas been cited as a oaue of savanna. It vas not observed in the Rupununi and probably is not extensive the re.

Wbere it does occur it may be a Resulting and a Maintaining Factor. A shallow pan irdieates a high vater table SJJd tree roots can often pierce a pisolithic laterite layer and possibly a t.ragipan. Tbe drainage in 99 the Cerr&dos seems to be good. The soils in general are very deficient in mineral.s. 'l'his may be a Predisposing Pactor or a

Resulting Factor. lutritional deficiency is an important factor in many savannas. It is not the sole cause as poorer soils are lmown to support forests. (The white sawds which are purer than powàered window glass support Wallaba (Eperua spp.) forest in parts of British

Gu.iana.) In the Rupununi savanna there 1s no closed nutrient cycle because of fire, leaching, and excessive drainage. Credence is given to Are na' the ory of Oligotrophic Scleromorphism. ( 195&, b, c.) • .

Studies of the causal factors of savannas are rare and difficult. Van der Haumen (l961a) has deduced :rrom the pollen record that the saV8.Illl88 of Northern South America were larger in the last

$lacial tban at present but there are signa that they may have started. expaming a gain vi thin the las t 3,000 years • Be dis eusses the climatic changes involved in the Quaternary, (196lb). Sauer (1944) arrived at a similar conclusion on savaona area using an anthropological approach. Hueck ( 1961) suggests trom Plant Geographical etudies tbat present savannaa are relicts of larger formations of the last glacial, am is suppOrted by Aubreville ( 1962).

Tbere st1U seems to be seme controvarsy over the Age and

Area Theory. Tbere is no dou.bt tbat "etdem1sm" will be reduced by better collection. Bates ( 1948) suggests that the low endem1sm of the savaDD4 fauna indicates the novelty of the savannas. This is supported by Vanzol1n1 (1963) from studies of disjunct distributions.

Tbe converse is suggested by Sick (1959) and Warming (1899) am by the observations of Ducke and Black (1953) on the disjuact distributions ot Rattlesnalœs 1n the sav8.1llla8. The distribution of plants should. . 100 prove to be a :f'ruitful avenue of research but at present the knovledge

of the flora is too 11mited to permit -very detailed studies.

A geomorpho1cgical. theory of savannas ru.· been postulated by

Ve1oso (1946) ar.d by Cole (1958, 1959, 1960). Cole suggests that

the Cerrado occurs mainly on flat senile landtorms, wbere the soils are not being reneved, and that tbe s1cpes support forests. This does not strictly apply in the Rupununi as the Pakarima ani Kanuku

Mountains have extensive areas of savumas on their s1opes.

Donselaar (1964) cJ.aims that the similarity betwen the mountain savannas am the white s&Di savannas of Suriname iniicate the ir former

~ geo1og1cal connection. This would have been during the Mesozoic when the Guiana Sbield started to rise l.eaving the Roraima sandstoœ plateau as isolated mountains. This tol.1ows tbe work of J'anshawe (1952) on the origine of the Guiaua flora. 101 APPRAIS.AL OF MIT!IOOS

Whenever a three dimens!onal mcdel has been constructed from a matr!x, the re have been conspicuous empt:[ spaces. The mo:lel is

a theoretical representation of the vegetation and probably haa space

for improbable comb !nations, (e .g. Wet gravel ridges), tbus accounting

for seme of tbe gaps. Bowever, in this case, 1 t is lilœly that some

of the gaps would be occup!ed by other types of vegetation. Ité swamp may be the true wet eai ot the continuum, B!llside savannas may be at the rich nutrient - good drainage part of the cœtinuum. Mur! Scrub may be an extreme low nutrient - sand type • These s 1tes should be sampled as they becœe more important in other areas, e • g. Palm savannas in the Llanos aJld Mu.ri Scrub in Suriname. An ioorease in the number ot staais used would also reduce tbe gaps • These quantitative phytosociological methcds vere devised :tor analysing well lmown vegetation in lforth America where gradient differences were so gradual tbat sopbieticated statistical procedures bad to be employed. The savannas are so 11 ttle lm.own that cruder methcds vould serve to accumulate intormatiœ on the main types •

Subsequently, sophiaticated methcds would be applicable. Presence lista tor stams would be usetul tor Constancy; aJJ4 P'requeney would initially be adequate for trees. The Point-quarter method vas derived for use 1n forests a:o:l 'Wbere the trees become scarce tb.is metbod becomes laborious. Long distance telescopic Range-f'iD:lers vould be usetul. Air photograph interpretation ot crown characteristics coupled vith grouDi control is beccming increaa!ngl;Y valuable as an &id

1n sucb ecological reaearcb.

Tbe aapling data are not preeented he re. Tbe vegetative 102 specimens bave not yet been related to the named specimens. It is probable that the "saplings" are not so much itdicative of regeneration

in tbe savanna but of a younger age class. It vould still be interest-

ins to aee how the canposition of the savanna is cbanging if the sapling

data sbould prove different trom the tree data. Coutinbo and hrri

( 1960) am Ferri ( 1960) show tbat germination is more common on the

better soils of new savannas be:f'ore 1mpover1sbment. J'rom this stu:ly

it follows that savannas may teai to the Campo Sujo type if given

suf:f'icient time.

1 " 103

Further avenues of research into savannas.. are suggested by this study. Savannas should be mapped Wi th greater acouracy. This could be pa.rt).y achieved from air photographe IIDi by collating existing maps. At the same time, a quantitative expression of tree density even vithout regard to the species 1 could be mede from the air photo­ grapbs • Tbese figures may be related to gross climatic etudies such as water deficit.

Very usetul intonetion cauld be calculated tram species Presence lista ot the savannu mapped. Bven a list of the Prevalent or Dominant species trom each savanna vould be of cCIIDSiderable value. Phytœociological

j studies, if they are based on quantitative methods so that savannas may be statiatically ccmparable, may yet yield much of the explanation. TlB y should be tried on the ettects of tire and grazing.

On a smaller scals, root etudies related to h}ldrology &nd trans­ piratiœ have provided the most s 1gn1f1cant vork to date. Tbese etudies vould be most interesting if made in other comparable areas. If burning could be controlled or quantified it would provide usef\11 information and belp clear up much controwrsy. Hovever, research studies not immediately related. to commercial gain are hampered more by lack of

:f\mds than by lack of ideas. 104 FOST ~RIPI'

Tbe people of the Rupununi savannas appear nei tber to need nor

want advice on how to run tbeir own aftairs. The .Amer1Ddians can

ob tain schoollng atld employtoent. The Ranchers bave a standard of

living in direct proportion to their efforts, while most Goverœental

empl.oyees work only temporarily in tbe Rupununi.

The Rupununi is a very poor area in which to malœ a living.

Jo amount of expertise, of whicb there is no lack, will produce a

panacea. .At present rancbing' appears to be the most suitable use

tor the Rupununi savanna. BcoD.OIIIic oœsid.erations preclude importation

of tertilisers • hom an ecological point ot view, tbe.re are tbree

items whicb may ameliorate the barsb enviroDmental ccalitions • Firstly1 tree planting, starting on a small ecale in the better sites. The

trees sbruld be Phreatophytes if possible. 'l'beae may reduce wiai and

tire, increase surtace soil, nutrients 8lld litter &Dd improve water

contUtiœ.s. If a crop can be gatbered trom the trees so mucb tbe better. The Caahew (Aœrcardium occW.entale) thri ves there alreedy.

'l'he Casbew pears ean be presened 8D:l the nuts used for oil or air­ exported as a ligbt veigbt-bigb coat commodi ty. Otber truit trees llllllY be as successtul. SeeODdzy, the rotation of Range Grazing would reduce the surface compactiœ by trampllng aD1 by rain, aggravated b:( tire am higb ins-olation. Tbirdly, small ecale water control by di tches and barrages voul.d reduce flooding and provide water tor the earzy part ot tbe dry season. Agricultural information coccerning the Rupununi and similar areas may be obtained f'rœ McCorkle (1952), McClung et al.(1957), 105 Gocde (1958) 1 Loxtœ et aL (1958), Stark et aL (1959), Rueck

(1961) 1 Rardy {1962) 1 Coimbra (1963) 1 Filb.o (l9Ç3) &Dl Freitas et al, (1963) • 106

Tbe vegetation of tbe liortbern Rupm1wù savaœa vas studied

quantitatively for the composition of tbe trees, shrubs, berbs 1 and

termitaria uaing tbe phytosoeiolOgieal approach of' the Wisconsin'

School. A total of 50 stanà.s 1 including 2 from lbini sa~ in

British Guiana and 3 from Trinidad savannas 1 were seleeted at random

and sampled • relative relative relative Importance Value (~hrequency + ~A>ensityf ;/Dominance) vas

calculated from ~he data derived by the Quarter-Method for trees,

saplings and termi taris. fo Frequency for shrubs and herbs was square calculated from occurrence in meter/quadrats. Tbese data were used

to determine the relationships of stands based on tbe phytosoeiological

similari ties of the herbe alone, by means of the Matrix Method. Three

axes vere derived; each axis having at i ts ends the most dissimilar

stams, vi th the intermedia te s tams ordinated between them • From

the three aas 1 a three dimens1onal model was construerted showing an abstract spatial relationship of' stands.

Quantitative values or the vegetation and of the envirœment vere then plotted cm this model. Obvioua patterns were talœn to

inUcate relationshipe 1 vlich were observed tor all trees, herbs 1 shrubs ani environmental factors tbat- bai been measured and that were plotted.

The patterns or 1nd1v1dual speeies all overlapped one anotb.er, but each was lmique and discrete. There vere strOJlg correlations between aspects ot the vegetation and envirœmental factors. (e .g. wter retaining capacity of the soil vith tree density and~ Cover; and between the occurrence of termi ta ria and level of soil nutrients.) From the 107 re sul ts 1 t appeared tbat tbe various types of vegetaticm tortaed a

c01'1tinuously varying series wbicb viradicated tbe ordinatiœ of the

starads as a continuum.

Tbe vegetation of tbe azoea waa domi.Dated by grasses ( Tracb.ypO&Op,

Ard.roeon. Paapall.J!!• Arist14!• ~· Meageetum, !r!s:irpt1a, Prmicum)

vitb sedges (J'!.!br1a~Yl1a, D1çhromepa. B\:Ynb08tYl1a, J!tm1choaREL

Stenop:QylJ8!, Cyperua, Scl!!i•> subd01Srwxt in places. Iutenpersed

among tbe grueea vere ~~&DY berbs (Q!J:!gt1a, hn'!ria, PolD!la, Cue1a,

MeJT!I!!f, Zvo1wlua). Trees (Çy.t!tella, Mxtlf9•M• Bmi!!i•, flupria, B2!!!1ich1a, Brxtbrcp;yl.og) were œwr dense &Dl vere reatricted to well

dl'81ned graval sites aD! rich moiat sites. Sb:rubs (btl-courea, Ramia,

Ccaptl'U§, Br.ytbrO!.Y;loo, CU!!J!!• )(yrt!C!u) vere usuaUy present. The

soils ve1"8 deep, senile 1 1mpowr1abed am ac1dic aand.a &Di gravels along

with clay and silt.

Tbe vegetation is cbaracterised by tbe :f'requent occurrence of

Xylopoclia storlng t.oo. all4 water1 by tbe large and sclerophyllous

l.eaves am th1ck bark of the treea, by the occurrence ot Pbreatopb;ytes

vith roots tapping tbe vater table 1 ao:l by peremiials am epb.emerals aDél intrequent annuals •

Tbe main teatures of tbe ecosystem ares Intense, pericdic

raintall of the short wet season, vith leacb1Dg1 sbeet-vash, erosion, su:rtace campaction 81:14 increued run-otr cauaing extensive floads

altemating vith the long drougbt vith t.Nque11t tires, removal of

humus IUld cover, 8Di the increue in viD41 inaolaticm aDd evaporation.

Tbe tend.Dology ot tbl savanzua bu been 41acûlsed; the teru

used being def1ned. A Uteratun reviev, 1Dclu41Di the important but

little knCM1 Brazillan work, bas been doculllllted vith copieu.a ntterences. e. 108 A description of the pbysical features of the area studied is given

together wi th brie t sections on the Geology, Soils, Geomorphology, Climate am Geog.raphy. The ·rauna, espec1ally the tsrmites is described with a section on the importance ot ire as a fBcter. As complete a list of species as possible. is presented for

the area; over 400 species belanging to 100 familles are noted.

A partial fioristic analysis is presented comparing :the Flora ot

the Rupununi with the Llanos 1 Campos Cerrados am other tropical savannas. Lista ot plants are presented for physiognomic, geo-

graph1cal am habit groups •

Conclud1ng sections are presented on the savanna ecosystem,

a critique Gf the iœthods used., EJCological obeervations am possible

avenues ot :f'Urther research.

,1 MCGill University Plant Ecology Research ' STAll> R!X.::ORD British Guiana Savannas Stam No: 109

Rame Loeallty: Map Sbeet: Air Photo: Date: Obeervers: Size o~ Stam: Size ot Area Sampl.ed: Vegetat10D Type So11 Type: Topography: Slope: Direct1œs

CoDi1 t1œ o~ Statd;

Open Grown Treee: rj, caver: Stumpe: l. 4. 10. 2. 5. u. 3. 6. 12.

So11 Data:

If K Ca Mg Org. Mat. Collo1ds W.H.C.

DEPTB Character1st1cs ( Color, Texture, Structure, etc.}

Parent Mater1al:

Map: · Termitaria:

r• 110 PBBSIJCE AID FBEQUIJCY PORM

Acisanthera Cyperus cusp~atus Aescbynomene .flavus Amasonia baspe.n Anarcardium luzulae Andropogon angus ta tua unioloides bico mis Cyrtopodium leucostacbyus Da villa selloanus Deamodium tener Dichromena Antœia Diodia Appunia Drosera Ariatida aetifolia Bchinolaena tine ta BJ.eocharis Axœopus anceps Bragrostis amabilis cbrysites gu.ianensis compresaus maypurenses Bacopa Briocaulon Bauhinia lr1oaeraa Boneria :lrythl'Olcylon Bolld1ch1a Bupe.torium Bucbnera elongata luphorbia palus tris Evolvulus Bulbostyl1s conifera l1mbr1styl1s terrugiDea Byrsonima coccolObaefolia llileacea crass1fbl1a monostachya st1pulacea Gal.act1a verbasc1fbl1a GyœD.opogon Cases rea Genipa Cassis cultrifolia Babenaria flexuosa Bellconia glamulosa Hibiscus h1sp1dula Birtella patellar1a Humir1s balaamifera tetrapbylla tlor1buttda v1scosa Hyptis Cassytha Ir:digof'era Ce reus ~pomoea Cipura Jacque mon ti& Cissampeloa Jussieua l1thospermifalia Clidemia aedo ides Clitor1a suffructicosa Coccoloba Lagenocarpus Coonarus Leptoco:ryphium Cordia Lippi a Coutouba Liaiantbus Crotalar1a maypu.renis Lorantbus retusa Mandevilla st1pular1a Meloch1a Curatella Mer remis Cu.rtia Mesosetum 111 P B B B B :1 C :& A li J) F R B Q U 1 1 C Y P 0 R M ( Continued)

Myrtaceae ( "Berber1e") Stylosanthes guayanensis ("pil.osa") biapidula leptunia Synganantbus Oxalis Tabebeu1a Paepalantbus Tibouchina Palicourea 'l'on ina Panicum cyanescens 1rachypogon laxum Turne ra parvi:f'olium Utricule.ria p1losum Waltberia stenodes Xyris Paspalum gardneriauum Zornia lachneum millegrana pulchellum ADDITIOUAL SPICIES: stellatum Pbaseolus grac111s lasiocarpus Pbyllantbus carol1niau niruri Pl.U81er1a Polygala adenophora angusti:f'olia langicaulis monticola Psidium Randia ~choapora barbata cephalotes cyperioides tenuis Bichardia Bosaceae Bou. pala Ruellia Sauvegesia Scbieckia Scbultesia Scleria bracteata hirtella tenella Sebastiaœ Setaria Sida Sipane a Spheoostigma Sporobolus Btachytarpbeta Stenopeyllus Stylosanthes angustifolia McGUl Unive:ra1,. ty

Plant Bcolcgy Besearch 112 British Guiana &momas Till, SAJILDI} AJI) 'l'.IRMITA1UA ..::011)8

Staticm ••· l 2 3 4 5 6 7 8 9 lO

1

1

'l.'DIIlTARIA '

; cova

J Data Form No. McGill University Plant Bcology Research TRIΠDATA SWIWti SHEET 113 British Guiana Savannas Stand lfumber:

1 # TOT # D.~ F.'/o F •. i> ~ SUM D.'/o DO. DJD PrS TRS DOM SAP SAP SOL T!tS F .TR TRS TRS lli>EX

AVE B.A.: 1 TRDS/ACBB DOM.JN:RI: 114 MATRIX FORM

SPIC DB SCORE V.A.t..U!

Adiantum Amasonia Aalropogon &DgUS tatus Aristida rad1ata Arist1da set1tolia Axoœp.~s chrys 1 tes Bauhinia Buclmera elongata Buclmera palustris Bulbostyl1s oollitera Bulbostylis paradoxa Byrsonima verbaso1tol1a Cactus Case area Cassytha Cipura Cissampelos Citrus Cl1tor1a Coccoloba Cœmarus Coutou.ba Cyperus Davilla Desmod1um Dichranena D1cxl:l.a Drosera Bcb.inolaeD& Bl.eocharis Bragrost1s glomerata lrythroxylou llupator1WD Bvolwlus Fern clhiber Galact1a G;ramopogon Habenaria liel1con1a Jaoquemont1a Juss1eua legenoearpus Leptocoryphium L1ppia L1s1antbus Loranthus 115 M A T R I X F 0 R M (Continued)

SPECIES SCORE VALtB

M'andevilla M!sosetwa Milky cl1mber Mimosa ~eae (Berber1s) Heptun1a Orchid {Green bog) Oxalls ~ Pallcourea Phyll.anthus carolln1ana n1rur1 Polygala ~t1tol1a long1caul1s Psidium Utricul.aria Ran:lia Ruellla Sauvegesia Schieeld.a Schultesia Sclerla ~;:r> Sebastlana Setaria S:14a Sipaœa Sphenostigma Sporobolus Stachytarpheta Stylosanthes Tibouchina Trachypogon Waltheria Zornia r•

116 PLANTS OF THE RUPUNUNI SAVANIAS

ACANTHACEAE Dipteracanthus angustifolius DOB Ruellia geminiflora BBK v.angustifolia (Nees) Griseba DOS

AimACEAE ALISMATACEAE Echinodorus N Lophotocarpus guayanensis (BBK) J .G.Sm. WEB

AMARABTACEAE Alternanthera dentatà WRB sessilis WRH Amaranthus gracilus WRB !resine elatior WR

AMARYLLJDACEAE Curculigo scorzoneraefolia WRH Hippeastrum solandrifolium WRH Hypoxis decumbens WRH

ANARCARDIAC.EAE Anarcad!!Dm occidentale L. MFT

ANNONACEAE M)T

APOCYNACEAE Mandevilla scabra DFS Plumeria inodora DFT ASCLEPIADACEAE Oxypetalum N

BIGNONIACEAE Pyroetegia venus ta (Ker-Gawl) Miers. N Tabebuia M:>T

BIXACEAE Bixa orellana N

BORAGINACEAE Cordia WRS Beliotropium fruticosum DRS 1nd1cum DRS

BURSERACEAI DRT BR:>MELIACEAE ... Ananas comosus MRS Brome lia kara tas L. - MRS Tillandsia flexuosa Sw. MRH BURMANliiACEAE 117 Burrœnnia WRH Gymnesiphon WRS

BtrroMACEAE N

CALLITRICHACEAE Callitriche N

CARYOPHYLLACEAE Polycarpaea corymbosa DOB

CACTACEAE Cereus DOS Melocactus neglyi DRS

CAPPARRIDACEAE

CHARACEAE Ni tella N

COCHLOSPERMACEAE Cochlospermum DRT

COMBRETACEAE Combretum MRT

COMMELINACEAE Commelina elegans Burm. F. WRll erecta L. WRB Dichorisamra hexamra (Aubl) Standl. WRH

COMPOSITAE Acanthospermum australe (L) Kuntze K>B brasilum M:>B Alomia DRH Baccharis DRH Bidens pilosus L. MRR Centratherum muticum (BBK) Lesa Clibadium surinamense L. WRH Elephantopua angu~~ifolius MRR Emilia sonchifoli• N Erechti ·~es h1erac.i.fol1a (L) Raf. N Eu pa torium amygdalinum Lam. l'l Ichthyothere t.erminalis ( S~reng) Blalœ DOR Orthopappus anguetifolius lSw) Gleason N Pectis elonga ta HBK K>R Riencourtia N Wedelia DRH Wulff'ia bac ca ta (L. f) Kun tze DRH rubens Aubl. DRH

CONNARACBAE Connarus incomptus WOT COIIVOLVULACEAE An ise :ta DIH 118 Evolvulus becasanus DPH sericeus DJH v. holosericeus DJIH Jacquemontia evolvuloides DFH . tamnif'olia Merremia aturensis (HBK) Hall f. Dli'B cissoides (J.am.) Hall f'. DPH juncea DFH Quamoclit coccinea N

CRASSULACBAE Tillaea N

CUCURBrrACEAE r• C'YPERACEAE Bulb os tylis conif'era ( Reichb.) Clarke C.B. DF lana ta (HBK) C.B. Clarlœ WR Cyperus articula ta WO cuspidatus HBK. WR fla vus (Vahl.) Nees WO haspan L. WF luzulae (L.) Retz WO platyphyllus WO simplex HBK. WO surinamensis WF unioloides R.Br. Jhchromena ciliata MF Eleocharis caribaea WO geniculata WO punctata. WO Fimbristylis annua DF ferruginea mile acea monostachya (L.) Hassk. 8padica Fuirena umbellata Rottb. WO Kyllinga odorats Lagenocarpus guian&nsis WF tremulus weigeltii Maris eus ( Cyperus) fla vus WF Rhynchospora armerioides Presl. MF barbata (Vahl) Kun th MF cephalotes (L) Vahl WO cyperoides longibractea ta subplumosa tenuis Link. MF Scleria bracteata MF hirtella Sw. melaleuca pterota " tenella Stenophyllus paradoxus (Spreng.) Standl. Dr DlLLEBIACKAE 119 Curatella americana L. WP"l' DaVilla aspera WFS '!\!!tracera N D:OOSERACBAE Drosera capillaris N ka1tulœns1s N

ERIOCAULACEAE Eriocaulon humboldt11 KUnth WOH tenui:tollum Kl. WOH Paepalanthus capillaceus lamarckii Kunth WOH' subtilis Miq. WOH Philodice hoffmanseggii Mart. WRH Syngonanthua umbellatus {Lam.) Rubl. WOH Tonina :tluviatile Aubl. WRH

EUPHORBIACEAE Caperonia palus tris {L) St. Hil N Croton trini talis Mill.ap. N Dalechampia affinis Muell Arg. N Jatropha urena L. B. Mabea N Manibot lf Euphorbia hypericifolia L. N Phyllan th us amarus WOH byssopi:toliua WBO carolinenaisWalt. · WOH niruri L. WRH stipularia WRH urinaria WOH ... Sebaatiana corniculata WRH linearif'olia Lanj • WRH

FLAOOURTIACEAE Casearea silvestria DFS Homalium racemosum MRT Ryania N

GENTIANACEAE Chelonanthus WRB Coutouba spicata Aubl. DFB Curtis tenuifolia (Aubl) Knobl. 'WOH Lisianthus WRB Nymphoides bumbold tianum (HBK) Kuntze B Schultesia benthamiana DC. WOH bracbyptera WOH subcrenata WOH GRAMIIIAE Andropogon angustatus{Presl.) Steud. D bicornis L. W. brevifolius Sw. GR.AMIDAE ( Continued) 120 Andropogon comenaatus HBK. sep. elongatus Hack. glome ra tus hirtif'lorus (Bees) Kunth. leucostachyus HBK. v. aubvillosus Hack. selloanus (Rack) Hack. D semiberbis (Bees) Kunth te:ner {Nees) Kun th virgatum Aristida capillacea Lam. D longifolia Trin. radia ta setifolia HBK. D. tincta T & R. ' Axonopua anceps {Mez) Bitche. D aetragalus aure us chrysi tes ( Steud.) KuhJman compressus (Sv.) Beauv. kaietulœnsis Swallen leptos tachyus ( ntigge) Hi tchc . suri~.f.s {Hochet.) Bemr. Brachy-aria eminii (Mez) Robyns Il Cenchrus echinatus N brownii DR Chloris polydactyle (L.) Sv. N DactyloU!tnium aegyptium ( L.) Richt. N Diectomis f'astigiata (Sv.) HBK K> Digitaria longiflora (Retz.) Pers. Bchinochloa colonum (L.) Link. Eehinolaena infiexa (Poir.) Chase WF Eleustne indice 1 llyonurus adustus (Trin.) Ekman N Bragrostis amabilis (L.) Wight. & Arn. Dll' glome rata guianensis Bithc. hy~Xloides maypurensis (HBK.) Steud. Gymoopogon f'ol1osua (Will4.) Bees K> Haclœlochloa granularia (L.) Ku.ntze liYlDinachne amplexicaulis ( Rudge) lfeea Icbftanthus acuminatus Svallen N axillaris (Nees) Hitchc • & Chase N Isacbne polygono1des (Lam.) Doell Iahaemum Leeraia hexandra Sv. W Leptoeorypbium lana tum Leptoehloa domtngensis (Jacq.) Trin. )feaosetum elongatum Mez D loliif'orme (Bochet.) Chase tenuifolium GRAMllŒAE ( Continued) 121 Panicum chloroticum Nees cyanescens Nees W aff' cvanescena Bees W elepbantipos hirtum Lam. 1axum Sw. maximum Jacq. W parvifolium Lam. W pilosum Sw. W rudgei stenodes Griseb. W trichoides Sw. versicolor Doell zizanioides RBK. Paspalum anceps arundinaceum carinatum d.ensum dilatatum gardnerianum Nees lachneum millegrana Schrad. orbiculatum aff. plica :..ulum aff. pictum poychae tum Mez polychaetum Mez pullchellum stellatum urville! Ryttillix granularis Sacciolepis myuros (Lam.) Chase strumosa (Presl.) Chase ap. nov. Setaria tenax (L.Rich.) Desv. Sporobolus cubensis Hitchc. Thrasya paspaloides petrosa (Trin.) Chase trin!tenais Mez Trichachne insularis (L.) Nees Tridens flaccida (Doell) Parodi Themeda arguens Trachypogon plumosus DF

GUTI'IFERAE Clusia N Symphanis N Vismia N

BAEM:>DORACEAE Schieckia orinocoensis WOH

HIPR:>CRATACEAE DRS

1 ~ HUMIRIACEAE 122 11umiria 'tialsamifera (Aubl) St. Bil. MRT v .guianensis (Benth) Cuatr. MRT :t'loribunda

HYDK>CHARITACEAE Anaeharis WRH

HYDK>PBYLLACEAE Hydrolea spinosa L. WRS

IRID.ACEAE Cipura paludosa .Aubl. WRH Sisyrhinchium alatum WRH marehioa WRH Spbenostigma gracile WRH

LABI.ATAE Hyptis americana MDS atrorubens MOS Marsipianthes ebamaedris WRH

LAURACEAE Cassytha americana MPB fili:t'ormis Mm

LEGtMINOSAE CABSALPINIOIDEAE Bauhinia bent-haminiana WRT Cassia cultrifolia Dl'B nexuosa PFR glandulosa nm hispidula DFH patellaria Ml'B tetrapbylla N v1scosa MFH swartz11 lf uniflora MFB K'rameria uniflora R Inga MRT MDI)SOJDEAE ~ Copaifera pubiflora Bentb. WRT Mimosa asperata l«>S pudiea WOS schrankioides WOS Beptunia WOS Swartzia R PAPILIONATAE Abrus pree a tori us lf Aesehyaomene filosa Mart. ex Benth. WOS histrix v. ineana (Vog.) Benth. panieulata Willd. ex Vog. pra tenais v. earibaea Rudd l.axa sensitiva LEGUMIR)SAB ( Continued) 123 Alysicarpus vaginalis DOS Bowdicbia virg111o1des HBK. W.PT Centrosema ft Clitoris guianensis DOH • rub1g1nosa Crotalaria goreensis DOH maypurensis re tus a stipularia Desmodium barbatum Benth (E. Oerst.) K:lH ct. canum incauum ..1 supinum v. angustU'olium Dioaloea R Eriosema violaaium MOB Galaatia jussieuana DES Indigo:f\!ra birsu ta MrR pasauorum Maab.aerium DOS Phaseolus grac111s DOII lasiocarpus Rbynahosia pbaseoloides DOS sehomburgkii Soemmeringa sempert'lorens Mart. WRH Stylosantbes angustitolla DPH guayanensis gracills hispidula Zornia dipbylla DPH

Lllfl'IBULARIACEAE Utrieularia benjaminiana N cornuta Michx. cuculla ta St • Hil. et Gerard follosa L. hydroaarpa longeeillata DC. myrioaista St B et G pusilla Vahl.

LILlACltAE Smilax N

U>GAIIACBAE Antœia ovata v. pilosa WP'l'

:WlWf.l'B.ACBAB Pboradeznron MrS Strutb.antbus N

LYTJ!RA CIAl Cupbea micrantha MrH

t' MALPIGHIACEAE 124 Ban1steria · N Byrsonima coccolobiaefolia coriacea MRT crassifolia MPT stip1lacea B verbascif'olia wrs v. villosa WFS Tetrapteris :t'imbripetala N

MALVACEAE Hibiscus WBS Pavonia speciosa WRS Sida eiliaris DPH corditolla glome rata jamaicensis linif'olia Urena lebata N

MARADTACEAE Maranta arundinacea L. N Oalathea N Thalia geniculata L. N

.MART'fHI.ACE.AB Oraniolaria annua N

MAYACACI!AE Msyaca longipes Mart. ex Seub. N

MBLASTOMAOBAB Acisanthera bivalvis (Aubl.) Oogn. WE crassipes (Naud,) Wurdack WEB nana Ule unifiora (Vahl.) Gleason WRR cf. uni:f'lora WEB Olidemia rubra (Aubl.) Mart. MRS Comolia lytbrarioides ( Steud) Baud. WOH villosa (Aubl.) Tr. sensu Gleasœ WOB Miconia albicans ( Sv. ) Tr • K>S eiliata MOS fallax DC. WRS ibaguensis (Bonpl.) T.r. WEB Pterogastra divaricata MRS Rbyncanthera grandi:f'lora wos Tibouchina aspera Aubl. WOS

MDISPERMAOEAE Cissampelos ovalifolia D.FS lk)JIMI.ACEAI

MJSAOEAE Beliconia humilia (Aubl.) Jacq. WRS (: B.psittacorum) MYRSIHACE.AE N 125

MYRTACE.AE Eugenia WRS Psidium WFS

Jn'CTAGIKACEAE Boerhaavia coccinea N

IIIMPRABAŒAI Cabomba B Nymphaea WHH

OCRN.ACEAE Sauvegesia erecta WOH Ouratea superba MRS

ONAGRACEAE Jussieua angustifolia WOS litbospermifolia nervosa peruviana potamogeton sedoides suf'f'ruc ticosa

O~lDACEAE Cyrtopodium aDderson.:U wœ Habenaria MJH Epider.drum Catasetum M)H Spirantbes

OXALlDACEAE Oxalls scbolliburgkiana DOH

PALMAE Bact;ois B Astrocaryum Jauri N Maur1t1a m1nor N Maximiliana regia B

PASSm.DBACEAE ~ssiflDra foetida N

PP.DALIACEAE Sesamum imicum N radiatum B

PBYTOLACCACEAE N

PIPERACEAE B R>LYGALACEAE 126 Bredmeyera MRR Polygala adenopbora DC • JI)H angust1fol1a R gal1o1des Po1r. MJH hygroph1la HBK. R leptoeaulls T et G long1caul1s HBK. DPH mont1cola HBK. WOS t1moutou HBK. R · sedo1des Benn. R subtilJ.s. DK. B Securidaca marg1nata Benth. WRS.

POLYGOKACEAE Coccoloba K>T Polygonum ..

POitrEDERIACIAE E1chorn1a cl"Sssipes 1 Heteranthera ren1form1s B et P. lf Reussis subovata R rotUDi1folla (L.f.} Caste llanos WOB

PORTULACACEAE Portulaca oleracea lf

PK>TEACKAE Boopala cauplleata K>T

RH.AMNACJUΠGouan1a velu tina R

ROSACEAE Chrysobalanus pellocarpa B v. 1caco. Birte l1a racemosa MP'l' L1canta B

RUBIACEAI Appunia tenu1tlora (Benth.) Book. f. DRS Borreria cap1 ta ta DJH ferrug1nea ocym1o1des (Burm.) DC. suaveolens vert1c1lla ta ( L.) G.J'. W. Meyer Cautar1a hexe.Jldra {Jacq.) K. Sebum. MBT Decllewda trut1cosa (Wtlld. ex R & S) IW.ntze WOH D1od1a hyssop1fol1a MOH oc1m1fol1a (Wtlld.) Brem ( D.rtg1da C et s) lrythal1s fruct1cosa R Gent pa ame ricana L. WOT Guettarda spruceana Muell. Are;. MRT M1 tracarpus tr1g1dus (W1lld.) K. Schum. M:>H scabrellus B RUBIACEAE ( Continued) 127 Oldenlandia laneifolia R Pagamea capitata Benth. MOT Palicourea rigida BBK. MFS Perama hirsuta Aubl. MRS RauUa formosa (Jacq.) K. Sebum. MFS Retiniphyllum schomburgkii (Benth.} Muell. Arg. • Richaldia scabra L. JfDS Sipaœa pratensia Aubl. WOR Spermococce tenior R Tocoyeœ DRS Uncaria guianensis Aubl. R

RUTACEAE lPagara pterota DRS

SAPIJI)ACIAJ Allopbyllus occidentalis MRS Cupania IRT PaulliDia teT

SCa>PHULAIU:.ACEAI Angelo nia sallcariitolia !f et B. R Bacopa flexilis WOH gratioloides (Benth.} lMwall retlexa (Benth.) ldwall salzmam11 (Benth.) ldwall sessillflora (Benth.) Edwall Buchnera e longa ta Dl.H palus tris {Aubl.) Spreng. WFll rosea Conobea aquatica Aubl. WBfl Liodernia crustacee (L) F.v.Muell WRH Scoparia dulcis L. WRH Vandellia diffusa L. WRB

SOLANACEAE Solanum americanum Mill N

S'IUCULIACEAE Buettneria involucrata DOR scabra Guazuma WRS Melochia hirsuta WOR villosa L. parvi folia HBK. Belicteres MRS Wal tberia ame ricana M:>S 11Xl1ca involucra ta SYMPIDCACEAE Symplocos R

THEACEAE Ternostroema R TURNERACEAE 128 Piriqueta cistoides WOll Vis co sa Turner& guianensis OOll ulmifolia OOll sp.

UMBBr..LIFERAE Hydrocotyle verticillata N

VERBENACEAE Aegiphila l«>S Amasonia campes tris MPR erecta MFH Lantana camara N Lippia schomburgkiana DFS Stachytarpheta cayennensis WOH elatior v. jenmani imica paradoxua ( Spreng • ) Standl.

VIOLACEAE HYbanthus ipecachuana R

VITACIAE Cissus erosa R

XYRIDACEAE Abolbod.a pulcbella H et B. WRH Xyris carolineana Walt. WRB: jupicai malmeana L .B. Smith paraensis Poepp. et Kunth

LYCOPODII21AE Lycopodium carolinianum v. meridionale (Underw. & Lloyd) Bessel cernuum L.

ISOETINAE Isoetes ovata Pfeiff. FILICINAE Adiantiopsis radiata {L.) Fee Adiantum pulverulentum L. terminatum Kunze Anemia villosa H. &B. ex Willd. Asplenium serro-denta tum Willd. Doryopteris collins {Raddi) J. Smith Dryopteris {Cyclosorus) gongyloides (Scbku.hr) Kuntze (Meniscum) permollis Maxon & Morton Lygodium micans Sturm venustum Swartz FILICmAE (Continued) 129 Poypoctium consimile laton decumanum Wllld • (Goniophlebium triseriale Swartz Scbizaea incurv~ta Scbkuhr pennula Swartz sp. n~. · Tricbomanes hostmannianum (Klotzscb) Kunze Blechnum cetarrucinum Pteridium aquilinum v. aracbnoidium ( Kaulf) Herter

MUSCI Cryptangium schomburgkiana P1ssideus kegelianue Octoblepbarum albidum Hedw. ,1 HEPATICAE Frullana exilis Lejeun1a repens Plagiocbila aûbplana Radula recubans

KEY

W- Wet M - Mesic D ·Dry N - Non-Savanna

(N • ....e .g. Aquatics, Swaœp Plants, Weeds, Cultivated Plants 1 Forest

Plants.)

R - Rare 0 .. Ocoasional F .. Frequent

H - Herb 8 .. Sbrub or woody T - Trees and Bushes herb 130 REFERENCES

Acocks, J.P.B. 1953 Veld Types of South A:f'rica.

Union S. Afr., De pt. Ag., Bot. Surv. 1 Me m. 28, 1·192 Adamson, A.M. 194o ·A second report on the termites of Trinidad BWl Trop. Ag. Trin. 17: 12-15.

Adjanohoun, E.J. 1962 Etude Phytosociologique des savanes de Basse Cote d'Ivoire (Savanes lagunaires} Vegetatio XI no. 1~, 1-38.

,1 Alvim, P de T. 1954 Teoria sobre a formacao dos campos cerrados Rev. Bras. Geogr. XV1: (4} 496-498. 1954

A1vim, P de T. and Araujo, W.A. 1952 Soi1 as an ecological factor 1n the development of vegetation in the Central Plateau of Brasil. Proc. 6th Int. Grassland Conf. Penn. State Coll. V.l, 610-él6

Anderson, H.'W. 1949 Does buming increase surface run-off'l J. Forest 47: 1, 54-57

Appun, C .F. 1871 Un ter den Tropen. Bri tisb Guiana. Jens Vol. 2.

Arens , K. 19 58a Consideracoes sobre as causas do xeromorfismo f'oliar. Bol. Fac. Fil. Cien Letr. U.S.P. 224 Bot. 15, 25·56

Arens, K. 1958b 0 Cerrado como vegetacao oligotrophica Fac. Fil. Cien Letr. U.S.P. 224, Bot. 15, 59-77

Arens, K. 1963 As plantas lenhosas dos campos cerre.dos como flora e.de.ptade. as deficiencies minerais do solo. Symposio sobre o cerrado, Univ. Se.o Paulo, 2e5-303

Arens, K., Ferri, M.G., & Coutinho, L.M. 1958c Papel do factor nutriciooal na economia d'agua de plantas do Cerrado. Reviste. de Biologie, 1 (3-4): 313-324.

Aristiguieta L. 1959 Plantas lndicadoras de incendies e.nuales. Bol. Soc. Venez. Cienc. Nat., 20 (94): 337-347: 1959 ..-

131 Aubreville, A. 1962 The formation of tropical savannas and quaternary glaciations. Adansonia, 2 { 1) : 16-84, 1962 .

Aubreville, A. 1961 Etude Ecologique des Principales :formations vegetales du Bresil. Nogent sur Marne, Seine, 286pp.

Aubreville, A. 1949 Climats, :foret et desertifiee tion de l'A t'rique tropicale. Soc. d'Editions Geo. Mar. et Col., 1 Vol. in 4, 351 pp.

Aubreville A. 1953 Les experiences de recons ti tutie r de la savane boisee en Cote d 1 Ivoire • Bois et :forets de Tropiques, 32: 4-10

Aubreville, A. 194 7 Les brousses secondaires en Afrique Equatoriale. Bcis et forets des Tropiques, 2: 24-~, 1947

Barran, C .N. 1956 Synopsis of geology between Pirara and Dadanawa Ranches, 1Npunun1 District, British Guiana. Rep. Geol. Surv. Dept. Georgetown, PP·9·

Bartlett, B.H. 1956 Fire, Primitive Agriculture and Grazing in the Tropics (in) 692-720 W.L. Thoœs ed. Man • s Role in changing the :face or the Earth. Chicago U.P.

Ba tes, H.W. · 1863 A naturalist on the River Amazon. London.

Ba tes, M. 1948 , CUma te and vegetation in the Villavicencio region of Eastern Colombia. Geog. Rev. 28: 555-574.

Beadle, lf .c . W. 194o Soil temperatures during forest f'ires and their eff'ect on the survival of vegetation. Jour. Ecol. 28: l.Bo-192

Beard, J .s. 1944 Climax vegetation in tropical America. Ecol. 25: 127-158

Beard J .S. 1946 The natural vegetation of Trinidad Oxf. For. Mem., 20

Beard, J .s. 1953 The Savanna vegetation of lorthern Tropical America Ecol. Monogr. 23: 149-215 i 132 Beard, J .S. 1949 Bras111an Campo Cerrado: P'ire climax or edaphic climaxt Geou ., Rev., 39: 664-666

Beard, J .s. 1955 The Classification of Tropical Vegetation Types. Ecol., 36z 89-100 Beebe, W. . 1925 Studies of a tropical Jungle Zoologies 6, 1-193

Belt, T. 1888 The Naturalist in Nicaragua. 2nd mt.

London, Bumpue 1 4o3pp •

Bennett, H.H. & Allison, R.V. 1928 The Soils of Cuba Trop. Plt. Res . Found, Washington.

Benoist, R. 1925 La vegetation de la Guyane Francaise Bull. Soc. Bot. France 72: lo66-1078

Bentham, G. 1839 Enumeratiœ of plants collected by Mr. Schomburgk. Ann. Mag. Nat. Hiet. to 1846: 105·111, 441-451, 427-438. Bews, J.W. 1917 Plant Successioo in the Thom Veldt. S. A~. J. Sei. 14: 153-172 BeYs, J .w. 1929 The World's grasses London, 4o8pp.

Bleackley, D. 19 57 Cyclostyled Report Dept. Geol. Surv. B.G.

Blydenstein, J. 1962 La sabana de Trachypogon del Alto Llano Bol. Soc. Venez, Cienc. Nat. 23 (102): 139-206

BouiUene, 1. 1925 Savanes equatoriales du Bas-Amazone (Bresil) C .-R. Associat. Franc. Avancement Sei., 48 Sessioos, 957-964 Liege.

Bouillene, R. 1926 Savanes equatoriales en Amerique du Sud Bull. Soc. Bot. Belg. 58: 217-223 133 Braun-Blanquet, J. 1932 Plant Sociology N .Y. The K!Graw-Rill, 439PP· Bray, J.R. 1960 The Composition of Savanna vegetation in Wisconsin. Ecolo 41: 785-790

Bray, J .R. &: Curtis, J .T. 1957 ... .An ordination of the Upland Forest Co111Duni ties of Southern Wisconsin Ecolo Monogr., 27: 325·349

Brock, S.E. 1963 Hunting in the Wilderness Hale, London, 208pp.

Budovski, G. 1958 The Ecological statua of Fire in Tropical American Lowlands. Actas 33, Congresso Internat. Americanistas, San Jose, Costa Rica., 1: 264-278

Budovski, G. 1956 Tropical Savanna. A sequence of forest felling and repeated burnings •• Turrialba • 6: 23-33 Burnett, J:,R. 1948 Crop Prodùction. 275-301 Tothill, J .D. ed. Agriculture in the Sudan, London' OUP. 974pp.

Burtt-Davy, J. 1938 The Classification of Tropical woody vegetation types Imp. For. Inst. Paper 13.

Carter, C .E. 1929 Lignotubers Aust. For.J. l2: 4

Charter, C.F. 1941 A Reconnaissance Survey of the Soils of British Honduras Govt. Printer, Trinidad 3lpp.

Chevalier, A. 1925 Epoques auxquelles des plantes cultivees et de mauvaises herbes pantrop• iques se sont repandues dans les pays chauds de 1' ensemble du globe. Rev. Bot. Appl., 5: 443-448

Chevalier, A. 1928 Sur l'origine des campes bresiliens et sur le role des Imperata dans la substitution des savanes aux forets tropicales. Acad. De Scienc. Paris C-R, 187: 997-999 134" Chevalier, A. 1929 Sur la degrad.atiœ des sols tropicaux causee par les feux de brousse et sur les fo:rmations vegetales regressives qui en sœt la consequence. C.R. de Seances de L•Ae. Sc. Paris, 188: BJJ-86

Chevalier, A. 1931 Le role l"ho!llle dana la dispersion des plantes tropicales. Echanges d • espaces entre l'Afrique tropicales et l' Amerique du Sud. Rev. Bot. App., 11 ( 120): 633-650.

Chevalier, A. 1933 Analogies et dissemblances entre les no res tropicales de 1' Ancien et du Jtouveau Monde C .R. Congr. Intern. Geogr. Paris, 1931. II 1933: 839-850

Coimbra R. de 0. 1963 Agricultura no Cerrado Symposio sobre o cerrado, Univ. Sao Paulo, 359·382

Cole, M.M. 1960 Cerrado, Caatinga and Pantanal Geogr. J. 76 (2)L 168-179

Cole, M.M. 1958 A Savana Brasileira Bol. Carioca de Geografia, Rio de J. 11: 5-52

Cole, M.M. 1959 The distribution and origin of the savanna vegetation vith particular reference to the "Campos Cerrado" of' Brazil. 18th Int. Geogr. Congr., Rio de J., 339·345

Cole, M.M. 1963 Vegetation Nomenclature and classification vith particu1ar reference to the Savannas. S. Afr. Geogr. J., 3-14

Corner 1 E .J.H. 1946 Suggestions :f'or !otanical Progress Bew Ph.yto~ 1 45: 185-~92 Corporacian Venezolana de Guayana 1961 Reconoc1miento agropecuario forestal del oriente de la Guayana Venezolana,

V. 3 1 Sabanas . Min. Ag. y Cria, Caracas, pp. 86

Cottam, G. and Curtis, J.T. 1956 The use of distance measures in phytosociological sampling Ecol., 37: 451-6o

Coutinho, L.M. & Ferri M.G. 196o Transpiracao e comportamento estometico de plantas permanentas de cerrado em campo de Mourao. Bol. Fac. Fil. Cien. Letr. Univ. Sao Paulo 247: Bot. 17: 119-130 135 Curtis, J .T. 1959 The Vegetat.ation of Wisconsin. Univ. Wise. Press. 657PP•

Curtis, J.T. &Cottam, G. 1962 Plant Ecology Workbook Burgess Publ. Co. Minneapolis, l93PP•

Curtis, J .T. & Partch, M.L. 1948 Effect of fire on the competition between blue grass aJJ1 certain prairie plants. Amer. Mid. Bat., 39: 437·43, 1948

Dansereau, P. 1947 The distribution and structure of Brazilian forests For. Chron. 23, 261-277

Davis, T.A.W. & Richards, P.W. 1933 The vegetation of Moraballi Cree-k B .G. J. Ecol. 21: 350-384

Davis, T .A.W. & Richards, P .W. 1934 The vegetation of ).Joraballi Creek, B.G. J. Ecol. 22 • lo6-155

Denevan, W.M. 1964 Observations on Savanna - Forest boundaries in Tropical America IGUJUIŒSCO Forest Boundary Symposium, Caracas, Mimeo. l4pp.

Denevan, W.M. 1961 The upland Pine Foreste of Nicaragua Univ. Calif. Publ. Geogr., 12: 251-320

Donselaar, J. Van. 1964 An Ecological and Phytogeographie Study of Nortbern Surinam Savannas. 114pp.

Donselaar, W.A.E. van 196-- Structure, root systems and periodici ty of savanna plants and vegetation in Northern Surinam The Veg. of Surinam. Vol. V7 Wentiat (Unpubl.)

Drummond, B. 1886 On the termite as the Tropical Analogue of the Barthworm Proc. Roy. Soc. Edin., l3: 137-146 ~' Ducke, A. & Black, G.A. 1953 Phytogeographie notes on the Brazilian Amazon An Acad. Bras. Cien. 25: 1-46

Duthie, D. W. 1939 The Soils ot British Guiana south of the 5th ~rallel, and of the Borth west District. Ag. J. B .G., 10: 173·193. 136 !gge11ng, w.J. 1947 Observations on the ecology of the Budongo Ra in Forest, Uganda. J. lko1. 34: 20-87

Egler, W.A. 1960 Contribuicoes ao conbecimento dos campos da Amazonia I · Os Campos do Ariramba. , Bol. lbs. Paraense Amilio Goeldi, N.S. Botanica, 4: 1-4o

Eiten, G. 1963 Habitat Flora of FaZenda Campininha, Sao Paulo, Brszil Simposio sob:rE. o Cerrado, ll:l. Uni v. S. P. 332 sp.

Erhart, 19 51 see Harris, 1961

Fanshawe, D .B. 1952 The vegetatiœ of British Guiana, a pre1im1nary review Imp. For. Inst. Oxford Inst. Paper 29, 96pp.

Fanshawe, D .F. & Swabey, C. 1948 Botanica1 and Ecologica1 exploration in Guiana. Timehri 28

:Ferri, M.G. 1943 Observacoes sobre Lagos Santa Ceres 21, Esc. Sup. Agric-Viscesa, Minas Gerais, 137-150

:Ferri, M.G. 1944 Transpiracao de plantas permanentes dos cerrados Bol. lac. Fil. Cien. Letr., 41: Botanica n. 4, 159-224

Ferri, M.G. 1955 Contr1bucao ao conhecimento da ecologie do cerrado e do caatinga. Botanica, 12, Fac. Fil. Cien. Letr., 195, U. Sao·Paulo, l-171

Ferri, M.G. 1960 Preliminary note on campo cerrado vegetation at Campo de Mourao, Parana. Bol. Fac. Fil., Cien. Letr., U. Sao Paulo, 247 (Bot. No. 17); 109-15, 1960

Ferri, M.G. 196la Aspects of the Soi1-Water-plant relationsbips 1n connection vi th some Brsz111an types of vegetation Proc. Abidjan Symp., 1959, Ul.'ŒSCO, 103-109

Ferri, M.G. 1961-b Problem of water relations ot some Brazillan vegetation types, vi th special consideration of the concepts of Xeromorphy and Xerophytism. Plant-vater Relations in Arid & Semi-Arid condition, Mardrid Symposium, œESCO, 1959, 191-197 • 137 Ferri, M.G. 1963 Ristorico dos trabalhos botanicos sobre o cerrado Symposio sobre o cerrado, Univ. Sao Paulo, 15-50

Ferri, M.G. & Coutinho, L.M. 1958 Contribucao ao conbecimento da ecologia~do cerrado. Bol. da U. Sao Paulo (Botanica 15) 103-148

Ferri, M.G. & Labouriau, L.G. 1952 Water Balance of plants :f"rom tbe "Caatinga". 1. Transpiration of some of the most frequent species of the "Caatinga" of Paulo Af'onso (Bahia) in the rainy season. Rev. Bras. de Biol. 12, 301·312.

Ferri, M.G. & Lamberti, A. 1960 Water economy of' plants in an enclosed area in Goiana (Pernambuco). Bol. Fac. Fil., Cien. Letr., U. Sao Paulo, 247 {Bot. No. 17): 133-45, 1960

Filho, O.A.G. 1963 Si1v1cultura no Cerrado Simposio sobre o cerrado, Univ. Sao Paulo. 383-lto8

Foldats, E. 1964 Mecbani.cal barriers in the soi1 of' Llano and the 1ntluence of these 1n the Pbyaiognoœy of the vegetatioo. !GU/UNESCO Symposium Ceracas

Follett-Smitb, R.R. 1930 Report of an investigation of the soils am of the mineral content of pasture grasses occuring at Waranama Ranch, Berbice R1wr. Ag. J.B.G., 3t 142-159

Follett-Smith, R.R. & Frampton, A. de K. 1935 Report on Agricultural conditions in the Rupununi district and Pakaraima Mountains. Ag. J. B,G. 6: 155-184

Fraser, B.C. 1946 Interior and Hinterland Bistrict Annual Report of the VJ!ter1ns.ryD1v. for 1945, B.G. Dept. Ag., 119-134.

Freitaljl, L.M.M. de, Mikkelsen, D.S., McClung, A.C., & Lott, W.L. 1963 Agriculture no cerrado Symposio sobre o cerrado, Univ. Sao Paulo, 323-358

Gilliard, E.R. 1962 On the breed.ing behavior ot the cock-of-the •rock. Bull. Aar. Mus. Nat. Rist., 127: 31-68

Gleason, H.A. 1913 The relation of Forest Distribution and Prairie Fires to the Middle West. Torreya 13: 73-181. 138 Gleason, H.A. 1932 The vegetational History of the Middle West Ann. Assac • Arœr. Ge or. , 12: 39-85

Gleason, H.A. & Cook, M.T. 1926 Plant Ecology of Porto Rico N.Y. Acad. ot Sei., Sei. Surv. of P.R. & the Virgin Islands, 7 (1): 1-96 • Gleason, B.A. 1932a The Progress of Botanical Exploration in Tropical South Arœrica. Bull. Torr. Bot. Club, 59: 21-28

Good, R.D. 1953 The gel")graphy ot the tlovering plants Longmans, London 452.

Goode, J.R. 1958 Livestock improvement & Range managerœnt at the Ebini 11vestock station (British Guiana) F.A.o. of the U.N. report 901, Rome, pp. 24

Graham, E.H. 1933 Flora of the Kartabo Region, British Guiana, Ann. Carnegie Mus • , 22: 17-292

Graham, V.E. 1964 Rupununi Plant List {Pers . Co!IID. )

Grasse, P.P. 1950 Termites et sols tropicaux Rev. Int. Bot. App1., 30: 549-554

Hammen, T. van der 196la Palynologische onderzaekingen in Zuid-Amerika Report M!eting of Naturawetensch. Studiekring Suriname en Ned·Anti11en, 18.2 .61, Utrecht, Not Publ.

Hammen, T. van der 1961b The Qua temary cllmatic changes of Northern Sou th America N. Y. Acad. Sei. 95: 676-683

!lardy, F. 1962 Fertillzation ~rob lems in the campo cerrado of E. central Brazil Turrialba, 12 (3): 128-133, 1962

Barris, W. V. 1961 'l'ermites, the ir recognition and control Longmans, London, 187pp.

Harrison, S.G. 1958 t Botanica1 Survey of Waranama am Ebini Intermediate savannas, British Guiana. Dept. of Ag. B.G., 1-34. 139 ·Harrison, S.G. 1963 Rupununi Plant List Nat. Mus. Wales, Cardiff, (Pers. CODJD.)

Hensel, J. 1923 "' J. Ag. Res. 23: 631-43

Hesse, P.R. 1955 Chemical and Physical study of the Soils of termite mounds in E. Africa J. Ecol., 43: 449-62

Heyligers, P.C. 1963 Vegetation and soil of a vhite sand savanna in Suriname The Veg. of Surinam III: 148pp.

Beyvard' P'. 1938 Soil temperature during forest tires 1n the long-leaf' pine region J. For., 36: 478-491 Bills, T.L. 196o A select annotated bibliography of the humid tropics. McGill Univ. Montreal, 238pp.

Hitchcock, A.S. 1922 Grasses of British Guiana Contr. U.S. Nat. Herb., 22: 432-514

Bi tchcock, A •S. 1936 Manuel of the grasses of' the West Im.ies U.S. Dept. Ag. Mise. Publ. 243. 439PP• Hitchcock, A.s. & Chase, A. 1917 Grasses of' the West Indies Contr. U.S. Nat. Herb., 18: 261-471 Boehne, F .c. 1939 Plantas e substancias vegetais, toxicais e medicinais Dept. Bot. do Est. S.P. Hoebne, F .C. 19lto tt/ Flora Brasilica Sao Paulo

Boldaway, F .G. 1933 The composition of' different regions of the mounds of' Eutermes r~xitiosus Hill. J. Counc. Sei. Im.ust. Res. 6: 16o-165

Boock, J. 196o The reclamation of dry sand savannah& in French Guiana Nouvelles de Guyane 10, also The Caribbean 14: 82-83 140

Roolœr, W.J. (ed.) 184o Geo Bentham Lond. J. of Bot. Vol. 1-5 Rough' w. 1926 Fire as an agent in human culture U.S. National Mus. Bull. 139: 1·270

Huber, J. 1900 Sur les campos de 1' Amazone inferieur et leur origine Act. Congr. Internat. Bot. Paris, 387-4oO

Rueck, IC. 1961 Distribution, ecology and economie importance of Cbaparrales in Venezuela Ber. Geobot. Inst. Rubel, 32: 192-203. 196o (1961)

Rumbold t, A. von & Bonpl.and, A. 1819 Persona! narrative of travels to the Equ1nocta1 Regions of America Trans. T. Ross, London, 7 Vols.

Ihering, H. von, 1907 A distribucao de campos e mattos do Brasil Rev. Mus. Pauliste, v.7, 125-178

ImThurn, Sir E.F. 1883 Arnong the Indiana of Guiana Lord on

Jackson, J .K. 1956 The Vegetation of the Imatong M:luntains, Sudan. J. Ecol., 44: 341-374

Jenman, G.S. 1882 Remarks on the aspects and nora of the Kaietuk savannah Timehri 1: 229-250

Jenman, G.S. 1888 The Savannahs of Guiana Hepr. from the Argosy, Jan 2lst, Feb.4th, March )rd.

Joachim, A.W.R. & Kandiah, s. 194o A comparison of so11s from termite mounds and adjacent land. Trop. Agriculturist, 95: 333

Johannesson, C.L. 1963 Savannas of Interior Honduras Ibero-Americana,· 46: l-16o (U. CaUf. Press)

Keay, R.W.J. 1947 Forest Vegetation in the Savannah Regions of Nigeria Fifth Empire For. Conf. 1-7 141

K'ellogg, C.E. & Davol, F.D. 1949 An explanatory study of so11 groups in the Belg1an Congo. !nat. Nat. Etude Agron. Congo Belge Serie Sei. 46: 1·73

K1111an, Ch. & Le mee, G. 19 56 Les Xerophytes: leur economie d'eau. Encycl,opedia of Plant Physiology, 3: 787-824

Knoppe, M.A. 196! Ecologica1 study on the termite fauna of the Jodensavanne region, Suriname. Stud. Fauna Suriname. (Unpubl.)

Kramer, P.J. 1956 Roots as absorbing organs. Encyclopedia of Plant Phys1o1ogy, 3: 188-214.

Kuhno1 tz-Lordat, G. 1939 La '1\trre Incendiee. Nimes, (France) l'ilition de la Maison Cairee Atelier Bruguier, 361pp.

Labouriau, L.E.G. 1963 Problemas de fisio1og1a eco1og1a dos cerrados. Simposio sobre o cerrado, Sao Paulo, 233-276

Lanjouw, J. 1936 Studies of the vegetation of the Suriname Savannahs. Nederlandscb Kruidkundig Arcbief., 46: 823-851.

Lanjouw, J. 1954 The vegetation and origin of the Suriname aavannas. Rapp. Comm. Cœgres. Internat. Bot. 8th, Paris 7&8: 45-48

Lauer, W. 1952 Humide und aride Jareszeiten in Afrika und Sud America.und ihre beziebung zu den vegetationsgurteln. Bonner Geogr. Abhaoolung. 9: 15-98

Lasser, T. 1955 Esbozo pre11minar sobre el origen de las f'ormaciones vegetales de nuestros Llanos./ Bol. Soc. Venez. Cienc. Nat., 16 (84): 173-200

Legge, S .P. & Bewson, R. 1962 Production of Beer Cattle at Bbini and Rupununi. Min. Ag. , For. and Lands, 17PP.

Linde man, J. C • 19 53 The Vegetation of the Coasta1 Region of Surname. The Veg. of Suriname I, I, 135PP· 142 Loxton, R.F., Rutheri'ord, G.K., Speetor, J., Jones, T.A. 1958 SoU and Land Use Survey&.. Imp. Coll. Trop. Ag., Trinidad. 2: 33PP•

Luetzelburg, P. von 1939 Die flanzengeographischen Verhaltnisse im Amazonas-gehiet. Ber. dtsch. bot. Geo. 57: 24762

Maguire, B. 1953 Mem. N.Y. Bot G;ns. 8 (2)

Maguire B. 1948 Plant Exploration in Guiana Bull. T·:)rrey Bot. Club, 75: 1-6

MCClung, A.C., Freitas, L.M.M. de, Gallo, J.R., Quinn, L.R., & Mott G.O. Preliminary f'ertili ty etudies on Campos Cerrados soUs in Brasil 1957 I.B.E.C. Res. Inst. 13.

McConnell, R.B. 1961 The Precambrien rocks of British Guiana. Timehri, 4o: 77-91

McCorkl.e, J. S. 19 52 Report on Ranch Study in the Rupununi vith Recommemation for improvement. Dept. of Ag. B .G. 1-29

McMillan, c. 1959 The Concept Vegetation and the comfortable Ecologist. Ecol., 4o: 488-490

McTurk, E.M. 1963 Karanambo, Rupununi, B.G. (Pers. Co!liD.)

Marshall, R.C. 1934 Physiograpby and Vegetation of Trinidad and Tobago Oxf'. For. Mem. No. 17: 7-56, 1934

Martius, K.F.P. von, et al, 184o-1906 Flora Brasiliensis Berlin, 15 vols. Martyn, E. B. 1931 A botanical survey of the Rupununi Development Company' s Ranch at Waranama, Berbice River. Ag. J. B.G. 4: 18-25

· Maycock, P .F. 1957 The phytosociology or Boreal-Conifer Hardwood Foreste of the Great Lakes Region. UDiv. Wise. Tbesis

,• 143

May cook, P .F. & Curtis, J. T. 1960 The phytosociology of Boreal-Coniter-Rardwood Foreste or the Great Lalœs Regions •

Ecol. Monogr. 1 30: 1·35 Meinzer, O.E. 1927 Plants as indicetors or ground weter U. S. Geol. Surv. Water Supply. Paper 577, 9 5pp.

Metcalfe, C.R. & Chalk, L. 1950 Anatomy of the Dicotyledons. Oxford, l500pp. Meyer, G.F .w. 1818 Primitiae Floree Essequeboensis

Michelmore, A.P .G. 1939 Observations on tropical African Grasslands.

J. Ecol. 1 27: 282-312

Mohr, E.C.J. & Baren, F.A. van 1954 Tropical soils Interscience Publ. N .Y. 498pp.

Morretes, B.t. de & Ferri, M.G. 1959 Contribuicao ao estuio da anatomie des :folhas de plantas do cerrado. BoL Fee. Fil. Cien. Letr., 243 (Bot. 1.6): 7-70

Murrovs, J .A. 1938 Vegetation,soils and termites. S. Afr. J. Sei., 35: 288

Myers, J .G. 1936b K'ew Bull. Mise . Inf. 10, 564

Myers, J .G. 1934 Observations on a journey from the mouth of the Amazon to Mt. Roraima and dawn the Cattle treil to Georgetown. Ag. J. B.G. 5: 86-100

~ers, J .G. 1936a Savannah and Forest Vegetation or the Interior Guiana Plateau J. Eco!., 24: 1.62-'184

Myers, J. G. 1933 Notes on the vegetation of Venez. Llanos. J. Ecol., 21: 335-349

Nye , P .H • 19 55 Some soi1 farming processes in the humid tropics J. Soil Sei., 6: 51-83 144 Oinonen, E. 1961 Anth111s as places for trees to grow, and as indicators of site history. Waldhygiene, 4(3/4): 67-79

Parsons, J .J. 1955 Miskito pine Savanna of Nicaragua and Honduras. AAAG, 45: 36-63

Patil, G.P. 1964 Statistics Dept., McGill Univ. Montreal

Pittier, B • 1926 Manuel de los plantes usuales de Venezuela Caracas, pp. 458

Pittier, H. 1947 Flora Venezolana Catalogo 2 vols. Caracas.

Plummer, F.G. 1912 Lightning in rtt1ation to :forest fires. U.S. Dept. of Ag. for. Surv. Bull., ,III 39PP•

Rachid, M. 1947 Transpiracao e sistemas subterraneos da vegetacao de verao nos 'Campo Cerrados' de Emas. Bol. Fac. Fil. Cien. Letr. &>, Univ. Sao Paulo, Botanica 5, l-14o

Rachid, - Edwards, M. 1956 Alguns dispositivos para protecao de plantas contra a seca e o fogo. Bol. Fac. Fil. Cien. Letr. 13

Ramia, M. 19 59 a Los Sabanas de Apure Min. de Ag. y Cria, Caracas, 137PP·

Ramia, M. 1959b La Vegetation como indicadora de los factores elima y Sue1o. Me morta de la Soc. de Ci en Nat. La Salle, 19 (53): 147-151

Ranzani, G. 1963 Solos do Cerrado Symposio sobre o cerrado, 51-92. Univ. Sao Paulo.

Rawitscber, R. 1942 Algumas nocoes sobre a transpiracao e o balanco d'agua de plantas Braeilieras. An Acad • Bras • Cien., 14 ( 1)

Ravitscher, F .K. 1948 The vater economy of the vegeta tion of the 'Campo Cerrados • in Sou them Brazil. J. Ecol., 36 (21): 237-268 145 Ravi tacher, F. 1950a Climax and Pseudoclimax vegetation in the tropics (South America) Proc. 7th Internat. Bot. Cong. Stockh;lm. 616-618.

Ravi tscher, F •. & Ferri, M.G. 1942 Observacoes sobre a metod~logia para o estudo da transpiracao Cuticular

em plantas brasileiros 1 especialmente em Cedrela fissilis. Bol. Fee. Fil. Cien. ~tr. Bot., 3: (U.S.P.) 113-138

Rawitscher, F.K. & Rachid, M. 1946 Troncos subterraneos de plantas Brasileiras. An Acad. Bras. Cien., 18: 261-28o.

Rawitscher, F. &Ferri, M.G., &Rachid, M. 1943 Prot'undidade dos solos e vegetacao em campos cerrados do Brasil Meri­ dionial. An Acad. Bras. Cien. 15 (4): 267-a94. Rio de J.

Richards, P.W. 1952 The Tropical Rain Forelt~ Cambridge, U.P., 450pp.

Richardson, W.D. 1963 Observations on the vegetation and Eoology of the Aripo Savannas, Trinidad. J. Ecol., 51 (2): 295-313

Rizzini, C. de T. 1963. A flora do cerrado Symposio sobre o cerrado Univ. Sao Paulo, 1.25-178

Rizzini, C. de T., & Heringer, E.P. 1962 Unierground organs of plants troœ some southem Braz111an savannas An Acad. Bras. Cien. 34 (2) 235-247

Rizzini, C •T. & Heringer, E. P. 1961 Underground organs of plants from some Southern Bras il savannas w1 th particular reference to the xylopodium. Phyton. 17 (1): 105-124.

Robinson, T .W. et al 1952 Symposium on Phreatophytes Transact. of the Amer. Geophys. Unioo. 33: 57·80

Roland, J.-CL, et Heydaclœr, F. 1963 Aspects de la vegetation dans la savane le Lamto (Cote d'Ivoire) Revue Generale de Botanique, 70 ( 834):: 6o 5-620 •

Roseveare, G.M. 1948 The Grasslands ot Latin Amrica. Imp. Bureau of Pasture & Field Crops, Aberystwyth.

Rudge, E. l8o5 Plantarum Guianae Rariorum 146 Rue, A.E. de la 1958 Sur l'origine naturelle probable de quelques savanes de la Guyane :francaise et de 1' Amazonie bresilienne. C.R. Soc. Biogeogr., 3o6: 50-53

Rutherf'ol'd, G.IC. 1958 Rupununi Plant List Dept. of Ag. B.G.

Sachs 1 C • 1879 Aus den Llanos Leipzig, Veit 369

Bands, W.A. 1961 lest structure and size distribution in the Genus Trinervitermes (Isoptera, Termitidae, Basutitermitinae.) .. N.ligeria termites. Insectes Sociaux 8 (2J: 177-100. 1961 Sandwtth, N.Y. 1929 Contribution to the South American flora Kew Bull., 1929 onwards •

Sauer, C.O. 1944 A geographie sketch of Early Man in America Geogr. Rev. 34: 529·573.

Sauer, C .o. 1950 Grass land Climax, Pire and Man. Jour. Range Management, 3: 16-21

Schimper, A.F.W. 1888 Plant geograpby upon a pbysiological basis Trans. W.R. Fischer 1903 Oxford

Schmeider, 0. 1927 The Pampa, a naturel or culturally induced grass land 'l Univ. Ca11f'. Publ. Geogr. 2: 255-270 '

Scbnell, R. 1952 Contribution a une etude phytosociologique et phytogeograpbique de l'Afrique occidentale. Mem. de l' I.F.A.N. 18: pp.234. Dakar

Schnell, R. 1945 Sur 1' orisine des savanes de la regions des Monts Nimba (Guinee Francaise) Bull. Soc. Bot. France 92: 249-251

Schnell, R. 1961 Le probleme des homologies phytogeographiques entre 1' Atrique et l'Amerique tropicales. Mem. du MUs. National. D'histoire Naturelle N.S. Bot., Xi, 2, 137-241

Schomburgk, R.B. 1922 Travels in British Guiana during the years 184o-1844 Leipzig 1848, Engl. Trans., Georgetown 1922 147 Schomburgk, R.H. et al 1848 Versuch einer Fauna und nora von British Guiana Leipzig.

Schomburgk, M.R. 1876 Botanical Reminiscences in British Guiana. Adelaide, 90PP•

Schubart, 0 • 19 59 Segunda contribuicao sobre o movimento da agua subteranea de Eœs• Pirassununga. Bol. P'ac. Fil. Cien. Letr. 243, Bot. 16: 71-84

Schubart, 0 & Rawi tscher, F. 1950 Notas sobre o movimento de agua subttN:'anea de Emas-Pirassununga. Bol. Fac. Fil. Cien. Letr. U.S.P. 109, Bot. 8: 69-73

Shantz, H.L. 1947 The use of fire as a tool 1n the management of the brush ranges 1n California Calit. Div. of For., 156pp.

Shantz, H.L. 1954 The place of grass lands 1n the earth 1 s cover of vegetation. Ecol., 35: 143-145

Sick, H. 1959 A formacao do cerrado 18th Internat. Geogr. Congr., Rio de J., 332-338

Siefritz, W. 1943 Plant lite of Cuba Ecol. Monogr., 13: 375-426

Sillans, R. 1959 Les savanes de 1 1 Afrique centrale. Paris Lechavalier, 423pp.

Simposio sobre o Cerrado 1963 Editora da Universidade de Sao Paulo. pp.424

Sinha, N.K.P. 1964 Geography Dept. Mc Gill Uni v. Mon treal (Pers • Com.)

Smith, A.c... 1945 The Vegetation of' the Guianas Chron. Bot. 16: 295-297

Smith, L.B. 1945 The Vegetation of' Brasil Chron. Bot. 16. 297-302

Spruce, R. 1908 Notes of a botanist on the Amazon and Andes ed.. A.R. Wallace, London, 2 vols. 148 Stark, J., Rutherford, G.K., Spector, J. 1959 Soil and' Land Use Surveys. Imp. Coll. Trop. Ag., Trinidad, 6: 24pp. Stewart. 0 .c. 1956 " Fire as the first great force employed by man W.L. Thomas ed. 'Man's role in changing the face of the earth'. Chicago U. Press, 115-133

Talœuchi, M. 196oa The structure of the Amazonien veg. I .. Savanna in Northera Amazon. J. Fac. Sei., Univ. Tokyo Sect. 3,7: 523·533

Talœuchi, M. 196ob A estructura da vegetacao na Amazonie. II - As savanna do norte da Amozon1a. Bol. Mus. Paraense Emilio Goelili, N.S. Botanica 7: 1-18

Tamaya, F. 1956 Contribucion al estudio de la flora Llanera Bol. Soc. Venez. Cienc. Nat. 17 (85):· 105-134

Tansley, A.G. & Chipp, T.F. 1936 Aims and methods in the study of Vegetation. London, 383pp.

Taylor, J.E. 1942 Termites Mounds used for manure. Farm and Forest, 3 ( 1) : 49

Trochain, J .L. 1954 Nomenclature et classification des milieux vegetaux en Afrique Noire Franctise 8th Congr. Internat. Bot. Sect. 7, 105-111

Troll, C. 1936 Termi tensavannen (in) Landerkundlicher Forschung, ~stschrift fur Norbert Krebs. Stuttgart (Englehorn) 275-312

Troll, C. 1950 (Pub. 1953) Savannentypen und das problem der primarsavannas. Int. Congr. Bot. Proc. 1: 670-675. Troll, c. 1956 Das Wasser als pflanzengeographischer faktor Bandb. d. Pflanzenphysiol. 3: 750-i86

Troll, C. 1959 Die tropischen graslander (Savannen) unter den e1nf1U:Ss von k11ma, boden, um wasser. 18th Int. Geogr. Congr. Rio de J. 302-307 149

Ve.nzolini, P .E. 1963 Problems fa unis ticos do cerre.do. Simposio sobre o cerrado, U. Sao Paulo, 305-321 .... Vares chi, V. 196o Observationes sobre le. tre.nspiracion de Arboles llaneros dure.nte la epoca de sequia. Est. Biol. Llanos, Ed. Venez Cien. Nat. 1: 39-45

Veloso, H.P. 1946 Consideracoes gerais sobre o vegetacao do este.do de Matto Grosso Mem. Inst. Oswaldo Cruz. 44: 579-6o3

Verdoom, F. (ed.) 1945 Plants and Plant Science in Latin America C:-Jron. Botan. 16 PP• 384

Vesey-Fi tzGerald, D. 1934 The great savannah district of British Guie.na. Trop. Ag. 11: 111·116

Vester, H. 194o Die areale und aree.l typen der Angiospermum-fe.milien • Bot. Archiv. 41

Viane., 0. & Araujo, W. 1946 Bol. Ag. Dept. Prod. Veg. Est. Minas, Gerais, 2 (11): 16-29

Viosca, P. 1931 Sponte.neous combustion in the marshes of Southern Louisiane.. Ecol. 12: 439-442

Voorde, P.K.J. van der 1957 Soil conditions of the ridge landscape and of the old coe.stal plain of Suriname BulL Lœ:l·ibauw-proefstation, Suriname, 74,210

Waddell, E.W. 1963 The Anthropic factor in a sa vanna environmen t. McG111 Univ. Thesis 207pp.

Waibel, L. 1948 Vegetation and Land Use of the Planalto Central of Brazil GeJgr. Rev. 38: 529-554

Wallace, A.R. 1878 Tropical nature and other essaya London.

Walter, H. 1939 Grasland, Savanne und Busch der arideren teile Atricas in ihrer okologischen Bedingtheit. J .b. Miss. Bot. 87: 750·86o Warming, E. 1~2 Lagoa Santa. Det. Kgl. Dansk Videnskabernes Selskab Skrif'ter 6 Raekke, Naturindensk og math afd.el. 6: 159: 488 ~ • Warming; E. 1899 Vegetation of Tropical America. Bot. Gaz., 27: 1-18

Weaver, J .E. 1954 North American Prairie Lincoln Nebr. 348pp.

Wild, H. 1952 The Vegetation of Southern Rhodesien termitaria Rhod. Ag. J. 49: 28o Willis, J .c. 1960 A dictionary of the f1owering plants and ferns. Cambridge U.P. 752pp.

ADDITIONAL REFERENCES

McConnel1, R. 1958 An Introduction to ~e fish fauna 0.1.~ B• G• Timehri 37 .

•

!••~ •